LIBRARIES SMITHSONIAN
m

INSTITUTION NOlinillSNl

NOliniliSNI NVINOSHimS

r*

LIBRARIES

NVINOSHIIWS S3ld^

tn

LI B R AR I ES^ SMITHSONIAN

INSTIT

INSTITUTION

^osv^

NOliniliSNI

NOliniliSNI

NVINOSHimS S3IH^

X
c/>
o

X

NVINOSHillNS^SS I B VH 3 ll^LI B RAR I ES^SMITHSONIAN INSTIT

INSTITUTION NOliniliSNI
z 1

NVINOSHilWS S31B^

NOliniliSNI

NVINOSHillNS SSIHVnan libraries SMITHSONIAN INSTIl

LIBRARIES SMITHSONIAN

- — ^ 2!

^NqC

NOliniliSNI

C/)

X

c/?
o
2

.. > S

NOliniliSNI NVINOSHlimS

S3IB)

NVINOSHili^S S3IBVBan

i

LIBRARIES SMITHSONIAN

1

BRARIES SMITHSONIAN INSTITUTION NOIinillSNI NVINOSHilWS S3IUVda

c/5 2 ^ 5 ^

liniliSNl"NVlN0SHilWS^S3 I a va a n'^LI B R AR I ES^SMITHSONIAN“lNSTITUTIO
, Z r- z ~

tn
>

0 7 :o

rn S fTj ^ m Xi^os»^

BRARIES SMITHSONIAN^INSTITUTION^NOliniliSNI "“NVINOSHill/yS S3 I H Vil 8 I

^ 1 O X O X

w I C/> C/) C/> I = ib I C/5

o
S

iiniiiSNi_NvmosHims^S3 1 ^ ve a n^ti b rar i es^smithsonian institutio

C/> - =: </7

:2

3-\ H Z H Q,

... <

OS

o -'w o 2*^ o

^ ^ ^ ^

3RARIES SMITHSONIAN INSTITUTION NOliOillSNI NVINOSHi^S S3iaV8a
z r- ^ ^ z r* „ 2:

w I I

liniliSNrNVINOSHilWS S3 I a VH a n“L! B RAR I Es‘^SMlTHSONiAN“lNSTlTUTI(

2 C^ 2

BRARIES SMITHSONIAN INSTITUTION N0linilXSNl”^NVIN0SHilIAIs”s3 I a Va 3

</> ^ ^ C/)

O

iJLni!iSNi"'NViN0SHiiws^S3 1 ava a n"*Li B RAR I es^smithson7an“institutic

^ X^JVASV^i^ ft)

BRARIES SMITHSONIAN"”lNSTITUTlON^NOIinillSNI*”NVINOSHillNS S3IBVaa
^ z * in z in

? s E ^ E

.. ? 'Kw z

2 fj) ■ -•'*'■ 2

finillCK.ll MWrKinCLi I IIAIC* C T I VJ W VJ O I T • a n n a n I r-

)7. 73

ANNAL^

OF

CARNEGIE MUSEUM

VOLUME 52

1983

PUBLISHED BY THE AUTHORITY OF THE
BOARD OF TRUSTEES OF THE CARNEGIE INSTITUTE
PITTSBURGH, PENNSYLVANIA
1983

MARY R. DAWSON, Acting Director {\/\/^?> to 5/31/83)
ROBERT M. WEST, Director {6/ to 12/31/83)

Publications Committee
C. J. McCOY, Chairman
LEONARD KRISHTALKA
GERARD McKIERNAN
JAMES B. RICHARDSON III
SUE A. THOMPSON

Editorial Staff

HUGH H. GENOWAYS, Editor
DUANE A. SCHLITTER, Associate Editor
STEPHEN L. WILLIAMS, Associate Editor
MARY ANN SCHMIDT, Technical Assistant

CONTENTS

Contents. iii

New Taxa v

Author Index vi

ARTICLE

i . The Caballo Muerto complex and its place in the Andean chronological

sequence. Thomas Pozorski 1

2. Studies on Nearctic Euchloe. Part 8. Distribution, ecology, and variation

of Euchloe olympia (Pieridae) populations. Harry K. Clench and Paul
A. Opler ■ 41

3. Further records of bats from the Central African Republic (Mammalia:

Chiroptera). J. E. Hill 55

4. New brachiopods from the Gilmore City Limestone (Mississippian) of

north-centrallowa. John L. Carter 59

5. Phenetic relationships within the Ciconiidae (Aves). D. Scott Wood . . 79

6. Systematics of Liophis reginae and L. williamsi (Serpentes, Colubridae),

with a description of a new species. James R. Dixon 113

7. Geographic and secondary sexual variation in the morphology of Eptes-

icusfuscus. Christopher D. Burnett 139

8. Resource exploitation in Amazonia: ethnoecological examples from four

populations. Eugene Parker, Darrell Posey, John Frechione, and Luiz
Francelino da Silva 163

9. Revision of the Wind River faunas, early Eocene of central Wyoming.

Part 3. Marsupialia. Leonard Krishtalka and Richard K. Stucky . . . 205

10. Paleocene and Eocene marsupials of North America. Leonard Krish-

talka and Richard K. Stucky 229

1 1 . The Chira beach ridges, sea level change, and the origins of maritime

economies on the Peruvian coast. James B. Richardson III 265

12. Landscape alteration and prehistoric human occupation on the north

coast of Peru. Daniel H. Sandweiss, Harold B. Rollins, and James B.
Richardson III 277

13. Principles of agrarian collapse in the Cordillera Negra, Peru. Michael

E, Moseley, Robert A. Feldman, Charles R. Ortloff, and Alfredo
Narvaez . 299

14. Results of the Alcoa Foundation-Suriname expeditions. VII. Records of

mammals from central and southern Suriname. Stephen L. Williams,

Hugh H. Genoways, and Jane A. Groen 329

15. Taxonomic review of Temminck’s trident bat, Aselliscus tricuspidatus

(Temminck, 1834) (Mammalia: Hipposideridae). Duane A. Schlitter,
Stephen L. Williams, and John Edwards Hill 337

16. South Asian middle Eocene Moeritheres (Mammalia: Tethyther-

ia). Robert M. West 359

17. Revision of the Wind River faunas, early Eocene of central Wyoming.

Part 4. The Tillodontia. Richard K. Stucky and Leonard
Krishtalka 375

18. Captorhinomorph “stem” reptiles from the Pennsylvanian coal-swamp

deposit of Linton, Ohio. Robert R. Reisz and Donald Baird 393

IV

NEW TAXA

DESCRIBED IN VOLUME 52

NEW GENERA

'\Annintodelphys, new genus, Mammalia, Marsupialia 221

'\Gerkispira, new genus, Brachiopoda, Spiriferida 72

'\Iowarhynchus, new genus, Brachiopoda, Rhynchonellida 66

NEW SPECIES AND SUBSPECIES

^Annintodelphys blacki, new species. Mammalia, Marsupialia 221

fArmintodelphys dawsoni, new species. Mammalia, Marsupialia 222

Aselliscus tricuspidatus koopmani, new subspecies. Mammalia, Chiroptera .... 353

Aselliscus tricuspidatus novaeguineae, new subspecies. Mammalia, Chiroptera 355

'\Gerkispira spinosa, new species, Brachiopoda, Spiriferida 74

'flowarhynchus mirandum, new species, Brachiopoda, Rhynchonellida 66

Liophis andinus, new species, Reptilia, Serpentes 129

'fSetigerites facetus, new species, Brachiopoda, Strophomenida 62

t Fossil taxa.

V

AUTHOR INDEX

Baird, D 393

Burnett, C. D 139

Carter, J. L 59

Clench, H. K 41

daSilva, L.F 163

Dixon, J. R 113

Feldman, R. A 299

Frechione, J 163

Genoways, H. H 329

Groen, J. A 329

Hill, J. E 55, 337

Krishtalka, L 205, 229, 375

Moseley, M. E 299

Narvaez, A 299

Opler, P. A 41

OrtlofF, C. R 299

Parker, E 163

Posey, D. A 163

Pozorski, T 1

Reisz, R. R 393

Richardson, J. B., Ill 265, 277

Rollins, H. B 277

Sandweiss, D. H 277

Schlitter, D. A 337

Stucky, R. K 205, 229, 375

West, R. M 359

Williams, S. L 329, 337

Wood, D. S. 79

/

) ^ ' t S

1

‘4.

I

^JC?-

>7. 7^

ISSN 0097-4463

ANNALS

0/ CARNEGIE MLISELIM

CARNEGIE MUSEUM OF NATURAL HISTORY

4400 FORBES AVENUE ® PITTSBURGH, PENNSYLVANIA 15213

VOLUME 52 30 MARCH 1983 ARTICLE 1

THE CABALLO MUERTO COMPLEX AND ITS PLACE
IN THE ANDEAN CHRONOLOGICAL SEQUENCE

Thomas Pozorski

Assistant Curator, Section of Man

Abstract

The Caballo Muerto complex in the Moche Valley, Peru, is a group of early ceramic
mounds, which were constructed sequentially between 1500 and 400 B.C. Recovered
ceramics show that most decoration and vessel forms cover the entire time span of the
complex. When the Caballo Muerto complex is compared with other early sites in Peru,
however, most types of ceramic decoration cannot be correlated with either the Initial
Period (1800-900 B.C.) or the Early Horizon (900-200 B.C.). Thus, the distinction
between the two periods is very vague. Attention is also paid to the fact that sites dating
to the Initial Period and the Early Horizon on the coast have radiocarbon dates several
hundred years earlier than supposedly comparable ones in the highlands. It is suggested
that the so-called Chavin or Early Horizon sites bearing modeled adobe friezes on the
coast actually predate the major constructions at Chavin de Huantar and Kotosh in the
highlands and that certain iconographic and architectural configurations originated on
the coast and diffused to the highlands. Basically, the Chavin phenomenon was not the
widespread unifying force that has been proposed by most investigators. Instead, through-
out early ceramic times (1800-200 B.C.) several regional polities developed and thrived
both on the coast and in the highlands.

The Caballo Muerto Complex

The complex of early ceramic mounds known as Caballo Muerto is
located on the north side of the Moche Valley on the north coast of
Peru. Investigations of the complex by the author lasted from July
1973 to December 1974. Though Caballo Muerto is known primarily
for the presence of adobe friezes at Huaca de los Reyes (Moseley and
Watanabe, 1974:154-161; Pozorski, 1975:211-251, 1976:63-92, 1980:

Submitted 30 April 1982.

5^\1HSu/V/^A
HH '' liio3

2

Annals of Carnegie Museum

VOL. 52

100-1 10, 1982:235-249), there are also chronological data, particularly
concerning ceramics, that need to be presented in order to fully ap-
preciate the significance of the complex.

Architectural Sequence

The complex consists of eight platform mounds of varying sizes that
are distributed over an area of 2 km^ (Fig. 1). Each mound is a corporate
labor structure that has, or probably once had, a pair of small parallel
wing structures extending out from its front face to form a large “U.”
The consistent “U” pattern plus the relatively large size of the mounds
suggests that all shared a similar nondomestic function. In addition to
the eight main mounds, the presence of a small niched room called the
Hall of the Niches just south of Huaca Cortada (Fig. 1) indicates the
possible existence of a ninth platform mound.

Among these mounds, however, certain architectural features differ
which suggest gradual chronological change. Five features were useful
in deriving a constructional sequence: 1) mound location, 2) main
mound size and configuration, 3) large mound-small mound associa-
tion, 4) mound orientation, and 5) entry pattern. The resultant seriation
(Table 1) indicates that the mounds cluster into three groups— Group
I consists of Huaca Cortada with the adjacent Hall of the Niches which
is probably part of a small mound, Huaca Herederos Chica, and Huaca
Herederos Grande; Group II contains Huaca de los Reyes, Huaca Cur-
aca, and Huaca San Carlos; and Group III is made up of Huaca La
Cruz and Huaca Guavalito.

There are two basic mound locations— on flat terrain or on a rocky
hillside. Huaca Herederos Grande, Huaca Herederos Chica, Huaca
Cortada, and Huaca Curaca are all built on flat terrain. Huaca de los
Reyes is basically constructed on flat terrain except that its western
half is situated on a natural flat-topped 6-m rise. The remaining three
mounds, Huaca San Carlos, Huaca La Cruz, and Huaca Guavalito are
all built on rocky hillsides in order to take advantage of the natural
base to add height to the principal part of each mound.

A second characteristic is the size and configuration of the main
mound. Huaca Herederos Grande and Huaca Cortada are large multi-
tiered platform mounds about equal in volume and height; the length
(defined as the distance along the principal site axis) of each mound is
greater than the width (perpendicular to the main site axis). Huaca
Curaca consists of two adjoining flat platforms of ascending height. In
a much more elaborate sense, the main construction of Huaca de los
Reyes has the same configuration (Pozorski, 1980:101-102, 1982:232-
235). On a much simpler level, Huaca San Carlos has the same layout.
For all three mounds, the length is again greater than the width, or

1983

PozoRSKi— Caballo Muerto Complex

3

Fig. 1. — Map of the Caballo Muerto Complex showing the distribution of its component
mounds.

very nearly the same. The final two sites, Huaca La Cruz and Huaca
Guavalito, have main mounds that are wider than they are long.

The phrase “large mound-small mound association” describes a re-
petitive pattern of one small mound situated laterally and adjacent to

Annals of Carnegie Museum

VOL. 52

UJ

b

+

c/3

.'I'S g

+ +

c

(U

■o

M

+ + +

X

■O 1

Paire

stair

case

+ + +

+

W

z

in

+

UJ

^ A C

E u

+ + +

— <u

e« BO

s?

c/3 C

+

||

+ + + + + + + +

u 1

z

o

+ + + + + +

"3

E

+ +

c/3

(u .JL -a

Larg

mult

tiere

+ +

(u "O — "2 «

Larg

moun

smal

mour

associ

tion

+ + + + +

c

Flal

terra;

+ + + + + +

rsi

■o

a

3

o

^ -O '=^ - O O

Cd_C3

S

Cortat
Niche
H. Ch
H. Gr
Reyes
Curac
San C

La Cr
Guav;

a

o

O

1983

PozoRSKi— Caballo Muerto Complex

5

a much larger mound. This association exists in three places within
Caballo Muerto: 1) between Huaca Herederos Chica (small mound)
and Huaca Herederos Grande (large mound), 2) between the Hall of
the Niches (small mound) and Huaca Cortada (large mound), and 3)
between the subsidiary mounds and the main mounds of Huaca de los
Reyes. Proximity and similar orientation make these associations sig-
nificant.

Mound orientation is another architectural seriation characteristic.
Five of the mounds have a common orientation, with the open end of
the “U” directed E10°N (magnetic north). Huaca de los Reyes, oriented
E5°N, varies slightly. Major orientation shifts occur in Huaca San Car-
los, opening E35°S, and Huaca Guavalito, at S10°E, respective 45° and
90° shifts from the predominant mound orientation.

The fifth characteristic is entry pattern, which is concerned with
access to the main mound as one passes through each site. Huaca
Herederos Grande, Huaca Cortada, Huaca Curaca, and Huaca San
Carlos all have central staircases, each located along the principal site
axis. Huaca Herederos Chica probably has or had a central staircase.
Huaca de los Reyes and Huaca La Cruz have both central staircases
and paired staircases bilaterally symmetrical with respect to the main
site axis. At Huaca de los Reyes, however, the central staircase is the
predominant access pattern whereas at Huaca La Cruz the opposite is
true. At Huaca Guavalito, the concept of central staircases appears to
have been abandoned altogether.

The resultant seriation of the Caballo Muerto mounds is supported
by the position of Huaca Curaca, a Group II mound, relative to Huaca
Cortada, a Group I mound. Huaca Curaca is situated such that it blocks
the entrance to the main plaza in front, or east, of Huaca Cortada,
indicating that Huaca Curaca is later in date.

The Caballo Muerto architectural seriation is based on the principal
constructions of each mound. There are indications of later architec-
tural additions and overlap in artifact assemblages which suggest ex-
tended use of some mounds during and/or after construction of other
mounds (Pozorski, 1976:114-120). This is evident for construction
phase 2 of Huaca Curaca which dates to the time of the main con-
struction of Huaca Guavalito. In addition, Luis Watanabe (personal
communication) recovered ceramics from the uppermost levels at Hua-
ca Herederos Chica which apparently date to the Group III occupation
of the complex.

Radiocarbon Dating

In general, the radiocarbon dates from each group of mounds support
the relative mound sequence (Table 2). Two dates from the same cul-

6

Annals of Carnegie Museum

VOL. 52

Table l.~ Radiocarbon dates for the Caballo Muerto complex.

Group

Caballo Muerto mound

Radiocarbon date

400 B.C.

III

Huaca Guavalito

Huaca Curaca, Phase 2

Huaca La Cruz

440 B.C. ± 70 (Tx-1939)

1400 B.C.

II

Huaca San Carlos

Huaca Curaca, Phase 1

Huaca de los Reyes, Phase 2
Huaca de los Reyes, Phase 1

850 B.C. ± 60 (Tx-2180)
1190 B.C. ± 60 (Tx-1973)
1360 B.C. ± 80 (Tx-1972)
1730 B.C. ± 80 (Tx-1974)

I

Huaca Herederos Chica

1090 B.C. ± 60 (Tx-1937)

Huaca Herederos Grande

Hall of the Niches

Huaca Cortada

1500 B.C. ± 70 (Tx-1938)

1800 B.C.

tural context at Huaca Herederos Chica, 1090 B.C. ± 60 (Tx-1937)
and 1500 B.C. ± 70 (Tx-1938), average 1300 B.C. From eonstruction
phase 1 at Huaca de los Reyes, four dates also averaging 1300 B.C. are
available from the same context— 850 B.C. ± 60 (Tx-2180), 1190
B.C. ± 60 (Tx-1973), 1360 B.C. ± 80 (Tx-1972), and 1730 B.C. ± 80
(Tx-1974). For the Group III mounds, a single date of 440 B.C. ± 70
(Tx-1939) was taken from a sealed floor of Huaca Guavalito. Though
there is some overlap in the dating of the Group I and Group II mounds,
it is clear that most of the mounds in these two groups predate 1000
B.C. The Group III mounds appear to date well after 1000 B.C., with
a terminal construction date of about 400 B.C.

Nonceramic Artifacts

The validity of the ordering of the architectural sequence is strength-
ened by nonceramic artifacts at some Group I and Group II mounds
which are similar to artifacts found at the early coastal site of Gramalote
that has six radiocarbon dates ranging from 1 590 to 1 100 B.C. (Pozorski
and Pozorski, 1979:418-421). These artifacts, found at Huaca Her-
ederos Chiea, Huaca Cortada, and Huaca de los Reyes, include jet
mirrors, stone bowl fragments, hammerstones, smooth stones, stone
paint palettes, and shell paint palettes. In addition, at the same three
mounds there are impressions of simple weaving, knitting, and straight-
paired twining preserved in adobe roofing fragments that are similar
to actual cloth specimens found at Gramalote (Conklin, 1974:77-92).

1983

PozoRSKi— Caballo Muerto Complex

7

NECKLESS OLLAS

3 4 5

10

11

12

13

14

15

Fig. 2.— Neckless olla rim profiles from Caballo Muerto.

Ceramics

The seriation of ceramic material is continuous when arranged ac-
cording to the stratigraphic evidence and the architectural sequence.
Absolute differences in ceramic traits are few, however, with most
differences in terms of degree rather than presence or absence.

Of a total of 10,240 sherds, the majority were found within archi-
tectural fill and fallen wall debris, though a significant portion was
associated with plastered floors. Thin (3-6 mm) coarse utilitarian ware,
both oxidized and reduced, makes up the bulk of the collection and
varies from 80% to 90% of the Group I assemblage to 65% to 70% for
the later mounds. Oxidized and reduced line ware constitutes the re-
mainder of the sample.

8

Annals of Carnegie Museum

VOL. 52

JARS

Fig. 3.— Jar and spout rim profiles from Caballo Muerto. Spout rims 1-3 are from stirrup
spout vessels and spout rim 4 is from a single-neck bottle.

Sixty-five percent of the 679 rim sherds found are made of coarse
ware. Vessel forms are limited to neckless ollas, short neck jars, open
bowls with straight or everted sides, stirrup spout bottles, and, rarely,
tall single-neck bottles (Figs. 2-4). Some partially restorable vessels
were encountered, indicating that most bottle forms and bowls were
flat-bottomed whereas neckless ollas and jars were generally round-
bottomed. Jars and bowls occur among types of both coarse ware and
fine ware. Neckless ollas are restricted to coarse ware types, whereas
stirrup spout bottles are virtually limited to fine ware types.

Pozorski—Caballo Muerto Complex

9

1983

BOWLS

Fig. 4. — Bowl rim profiles from Caballo Muerto.

Neckless ollas predominate throughout, but are especially frequent
at the earlier mounds. Neckless olla rim 8 (Fig. 2) seems chronologically
significant because, though present at all of the mounds, it occurs only
in the first construction phase at Huaca Guavalito. Neckless olla rim
form 9 (Fig. 2) is present only at Huaca de los Reyes and later mounds.
Diagnostic jar rim forms include numbers 3 and 7 (Fig. 3) for the early
part of the sequence and numbers 9 and 10 for the end— within con-
struction phase 2 of Huaca Curaca and at Huaca Guavalito.

Of the numerically significant bowl rims, the absence of form 6 (Fig.
4) separates Huaca Guavalito and phase 2 of Huaca Curaca from the
other mounds. Stirrup spout fragments (Fig. 14) were present at all

10

Annals of Carnegie Museum

VOL. 52

Fig. 5. — Incised and crosshatched sherds from Caballo Muerto: a-c, e are from Huaca
Herederos Chica, d is from Huaca Guavalito, / and h-i are from construction phase 2
of Huaca de los Reyes and g is from construction phase 2 of Huaca Curaca.

sites, and much more common than definable stirrup spout rim forms
(Fig. 3). The only two definite single-neck bottle rims found within the
entire complex came from construction phase 2 of Huaca Curaca.
Specific rim forms are not numerous enough to be considered diagnostic
in themselves, but several fragments of stirrup spout arcs indicate that
all examples are without decorative modeling and are trapezoid or
rectangular in profile.

A total of 856 decorated sherds were recovered, 70% of which are
made of fine ware. From the beginning of the sequence, there are
numerous modes of decoration, most of which persist throughout the

Table ?>.— Distribution of Caballo Muerto ceramic decorative modes 1 through 7 by number and percentage.

1983

PozoRSKi—- Caballo Muerto Complex

11

0

'5>

O' ^

^0

On

0

NO

0

1 ! ^ 1

Cl.

CC- 1 1

Cl

1— H

1

^ '

' —

ry^

ro

rs <N

(N

10

>

^ s ^

3^ Co

#

vP

NO

frj

^ 'O

VO

c

0

GO O) (N j

Cl.

! CnJ 00 1

Cl

rs|

i

NO

^ —

*0

fO fO o^

00 ro

On

<N

>0

0^

'9

#

0

. ■ ^ .

<N

(N

<0)

1 1 0 1

Ci

1 Cl 1 1

1

1

NO

' — '

' —

fN

fN

•o

m

00

/ — ^

#

^ ^ 0 ^
O' 0^ ’— 1 0^

^ ^

-5 2.. -P

O' 'T O' ^

VO

#

0

#

#

10

0

0

<N

Tf <N CM 0

Cl

<N m 0 ^

d

(N

r—i

' — ^

'. — '

ro t" VO ^

(\

VO —1 CO

NO

00

00

-.1 O)

"'I

0

''~l

NO

Tl-

rr)

fVi

Cl

#

#

0

1

1 1 1 1

CT' 0^

' C ' C

t\

r-

1

0

10

0

•N

cn

NO

^ ^

-.o'

^ ^ ^

^ Cp

NO

^ ^

# (N

^ ^ 00

0

Ci

w 2^ 0 1

(N ir>

1 d

VO ^

—

VO '*=t

>

(N (N

>0

VO ^ 00

0

<N

*0

10

<N

0

^ ^ -3

Qo ^ H oo ^

rr^ OO —I ^ — I

I (N ^ ^ — I ^

(U

«J s =

■g 2
c-guo
5 iz K I

<N

cd cd

U ^

(/3 (/3

OJ (U

<U <U J3 3

Pii Pi U

N

S g

20

■5 I

2 gS

u

^ k ^

00

C C
o o

■2

. o

e G

o G
a

2 T3

o <u
G G
G O
CU N

2

o

G

G

a c

t3 .2

<U crt

§'i

N .S

<N IT) VO

7) Incised raised bands.

12

Annals of Carnegie Museum

VOL. 52

chronological sequence, with new modes simply added to the repertoire
(Tables 3-7). For the Group I mounds, there are 13 modes of deco-
ration: 1) fine-line incision (1 mm or less wide), 2) broad-line incision
(more than 1 mm wide), both of which are executed in wet paste forming
geometric patterns (Figs. 5a-b, lOa-b, d, f-h, llh, 15a-d); 3) circular
and diagonal punctation, both unzoned and zoned by incised lines (Figs.
8c-e, 9b-g, 16a-d); 4) crosshatching (Figs. 5c, e, 10c, lla-b, i); 5)
combing, a type of decoration consisting of a series of parallel, non-
overlapping incised lines made by use of an implement with six to nine
teeth; 6) raised ribs or bands, usually marked by incisions executed
transversely or obliquely to the length of the band (Fig. 1 2d); 7) mod-
eling, probably part of figurines or figured vessels; 8) pattern burnishing,
usually in the form of parallel lines; 9) applique bumps, measuring a
few millimeters in each direction and usually bearing incision (Fig.
12h); 10) post-fired paint within and outside incised lines; 1 1) rim or
lip incision (Figs. 13e, 17a); 12) square rim castellation; and 13) black
zoning, probably consisting of graphite. There are also numerous ex-
amples of sherds bearing combinations of the above techniques.

For mounds of Groups II and III, all forms of decoration except rim
castellation continue, usually in continuous succession, but with certain
frequency fluctuations (Figs. 5d, f-i, 6c, 7d-g, 8a-b, f-h, 9a, h, lOe,

1 Ic-g, 12a-c, e-g, 13a-d, f, 17b-e, 18a-d, 19a-d). A significant chron-
ological marker separating these latter two groups from Group I is the
use of a black, probably graphite, paint. It is used both as a filler of
incision, often in combination with punctation or raised bands (Figs.
6a-b, 2 1 e), and as black bands or zones painted on flat surfaces, usually
demarcated by incised lines (Figs. 6d-f, 7a-c, e-g, 20a-e, 21a-b). De-
spite the presence of two black zoned sherds at Huaca Herederos Grande,
these techniques statistically appear to be definite time markers, and
are exclusively associated with two types of reddish brown burnished
fine ware. The use of graphite or manganese as a slip entirely covering
one or both surfaces of a vessel (Figs. 6g-h, 21c-d) is especially abun-
dant at Huaca Guavalito and therefore further distinguishes this mound i
from other mounds within the complex.

The end of the mound sequence at Caballo Muerto is distinguished
by four additional decorative modes. Plain and dentate rocker stamping
(Fig. 1 3g-h) occurs only at Huaca Guavalito and in construction phase

2 of Huaca Curaca. One sherd recovered from Huaca Guavalito appears
to be “brushed,” bearing a dense cluster of multiple incisions that
generally run in the same direction, but frequently overlap. Cane- i
stamped and incised circles (Fig. 22a-g), usually about 2 cm in di-
ameter, exist in small numbers at the remodeled Huaca Curaca and in
greater quantity at Huaca Guavalito. A lone example of stamped circles
is seen in Huaca de los Reyes, but these are much smaller (4 mm in

Table A.— Distribution of Caballo Muerto ceramic decorative modes 8 through 14 by number and percentage.

1983

PozoRSKi— Caballo Muerto Complex

13

# o

i i

I I i I I I I i i I

! I I I i

I I I

I I I I I I ! I I I

I I I ! I I

I I I I I I I I I I

00

1 I

*0 I I

(N I I

c/i -p I

l^uo

5

U Z DC X

On

rsi

I 1 2

o o
o o

2 I I

o o
o o

2 I !

fVN ro
'v, rn

I !

I ! I I I

— ' <N ^ .

^ 5/) c/5 r ) o

7^ w u ^ CO

O Cii Di u

I

u

3 01 2
U J O

I I

^ s

s I

(/>

a> a>

C C

TS xi

(U OJ
t/3 V)

o o
C C!

P. P

>>

c c .

cC oh
p p d

e e-^

O O c

s

o o o

03 d o

^ ^ ^
'% '% '% % 'I

^3 ^ T3 T3

o D <u a> (L)

(/3 c/5 c/3 1/5 c/5

T3 'O T3 XJ T5
i> C.) C.)

C/5 5/5 C/D C/3 5/5

‘o ‘G 'C *5 *o

1 3) Incised raised bands with graphite zoning.

14) Figurine or modeling fragments.

Table 5. — Distribution of Caballo Muerto ceramic decorative modes 15 through 21 by number and percentage.

14

Annals of Carnegie Museum

VOL. 52

I I

I I I I I I

I I I I I I I I I I

I I z: I

I I I I I

I I I

I I I I I

:d

<u
T3

C3 o £

uzxx

I Z! I I

z: i I

1^11

I ^ i I

o o

«N CM
o-i

m

cd cd

c/} !/i r ) u

<D <D ^ C3

^ ^ C b:

<u <u S =*

DiJ c/5 U

I I

I I

I I I

I I

U

N

2 2
u ^

o

<N

s

00

S

O

o

22

1 ^

x)

(N

(N

Os

c

d .2

CIS </5

<u ^
^ is ^

a'$

cv

&

ll

1 1

c/5 {/3

*G ’u

^ a
^ a

l-g s

s| i

o o 2

03 o cx

^ ^ ^
'flj '(U '(U
3 ;3 3

• S'. S'. S'
a aa
cx a a
< < <

Table 6.— Distribution of Caballo Muerto ceramic decorative modes 22 through 28 by number and percentage.

1983

PozoRSKi— Caballo Muerto Complex

15

I I I

I I

I I

I I I 2

I I

VO

<N

0\ oo
90 (N

VO

o

^ 1

1 r- r-

1 fO 1

1

1 1

o

1 f— 1 y~-i

22

1 m 1

22

1 1

^ '

'' —

<N

C ^

•N

<N

<N

<N

Q

I I I I I

I I

o

(N

27) Combing with graphite zoning.

28) Rocker stamping and incision.

Table 1 . — Distribution of Caballo Muerto ceramic decorative modes 29 through 34 by number and percentage.

16

Annals of Carnegie Museum

VOL. 52

I I I I I I I I I I

I I I I I

I I I I I II

<u

« s|

■gj-- 2

o •

UZXX

U O

I I 2

I I I I I I I I I I

I I I I I I I I I I

o ^
o

^ <N

I I I I I I I I I I

o o
o o

o o
o o

o o

Cn, ^ —I

2 I I

I I

(/i t/1

O (U

<U <V)

C3 c3

u

o

N

2

CJ

J O

o .2
a: u

"u

u

'o

_o

C oi)

C s

-C

cl

o 2

U P3

1983

PozoRSKi— Caballo Muerto Complex

17

Fig. 6. — Decorated sherds from Caballo Muerto: a is from Huaca La Cruz, b-d and g~
h are from Huaca Guavalito, e is from construction phase 2 of Huaca Curaca, and / is
from construction phase 2 of Huaca de los Reyes. Sherds a-b have graphite within incised
decoration, c is decorated with incision and punctation which is covered with an exterior

graphite slip, d-f are decorated with painted graphite bands, and g-h have stamped
circular decoration.

diameter) than the normal sized circles present at the other two mounds.
The larger circles at Huaca Curaca and Huaca Guavalito occasionally
contain either a small central circular punctate or a small central con-
centric circle (Fig, 6g-h). The circular designs are usually associated
with the fine ware type that is covered completely with graphite or
manganese, but are also present on a reddish brown fine ware. Broad-
line incision executed upon the graphite covered fine ware represents
the only possibility of naturalistic designs within the entire Caballo

18

Annals of Carnegie Museum

VOL. 52

Fig. 7. — Decorated sherds from Caballo Muerto: a is from Huaca Guavalito, b-e and g
are from construction phase 2 of Huaca de los Reyes, and /is from construction phase
1 of Huaca Curaca. Sherds a-c and e~g have painted zones and bands of graphite, g
also has zoned punctation, and d has incised decoration.

Muerto complex. One example may represent a claw whereas a few
others might once have been parts of feline motifs.

In comparing the Caballo Muerto ceramic assemblage with that of
Gramalote, the correspondence is very close, especially for Groups I
and II. The primary vessel form at Gramalote is the neckless olla with
lesser quantities of jars and bottles. Decoration consists of incision,
zoned punctation, and decorated raised ribs or bands.

1983

PozoRSKi— Caballo Muerto Complex

19

g

0 2 cm

Fig. 8. — Sherds with punctation and zoned punctation from Cabailo Muerto: a is from
Huaca La Craz, b is from construction phase 2 of Huaca de los Reyes, c-e are from
Huaca Herederos Chica, / is from construction phase 1 of Huaca de los Reyes and g-h
are from Huaca Guavalito. Sherds g-h were probably decorated using a multi-pronged
instrument.

Initial Period Versus Early Horizon

Up to now, the terms “Initial Period” and “Early Horizon,” using
the Rowe (1960:627-631) terminology, have been avoided during this
discussion of Cabailo Muerto, with preference given instead to the more
general term “early ceramic.” The reason for this avoidance is simple —
the distinction between the Initial Period and the Early Horizon is not
a clear one. In theory, the distinction seems obvious. Using the Ica

20

Annals of Carnegie Museum

VOL. 52

Fig. 9.-— Sherds bearing punctation and zoned punctation from Caballo Muerto: a and
h are from construction phase 2 of Huaca de los Reyes, b and d-g are from Huaca
Herederos Chica, and c is from the Hall of the Niches.

Valley to set up a master sequence, Rowe (1962(2:49) designated the
Initial Period as the time of the introduction of pottery in lea. The
Early Horizon begins with the first appearance of Chavin influence at
lea. It is characterized by the Chavin style and lasts until polychrome
slip painting replaces resin painting in that valley (Rowe, 1960:628).

However, as Lumbreras (1974(2:49) has pointed out, just because the
Chavin cult is predominant in the Early Horizon, this does not mean
that contemporary local cultures fold up and disappear. Several areas
of Peru, especially southern Peru, were never touched by Chavin in-
fluence (Willey, 1951:137) despite the exaggerated claims of Carrion
Cachot (1948:^166) and Tello (1943:159, 1960:36-41). In the areas of

1983

PozoRSKi— Caballo Muerto Complex

21

Fig. 10. — Incised sherds from Caballo Muerto; a-d and /-/? are from Huaca Herederos
Chica and e is from construction phase 2 of Huaca de los Reyes.

postulated Chavin influence, distinctions become blurred, especially
with respect to ceramic decoration, because many techniques used
during what has been defined as the Early Horizon were also employed
in the time range designated Initial Period (Panning, 1967:101).

First and foremost, it must be remembered that Chavin is essentially
an art style, recognizable and distinguishable from other ancient Pe-
ruvian art styles. It is when one tries to relate the art style to cultural
mechanisms that problems arise. If one adheres to the definition pro-
posed by Willey (1951:138) of the Chavin style as “identical to, or
closely resembling the designs of the stone carvings of the type site,
Chavin de Huantar,” then he is on fairly safe ground. If a stylistically
definable Chavin figure, such as a feline, a bird, or a serpent is present,
then one can assume that the culture that created such an artifact was
in some way in contact with the Chavin phenomenon. Carrying this

22

Annals of Carnegie Museum

VOL. 52

Fig. 11. — Incised and crosshatched decoration from Caballo Muerto: a-b and h-j are
from the Hall of the Niches, c-f are from construction phase 2 of Huaca de los Reyes,
and g is from construction phase 1 of Huaca de los Reyes.

process a step further, artifacts and other materials directly associated
with Chavin art are then assumed to be closely connected with the
Chavin phenomenon, especially if no local antecedents appear to be
present at the site in question. Further, if artifacts associated with
definite Chavin style objects are discovered at other sites without true
Chavin style associations, then it is often assumed that these contem-
porary objects that are not stylistically Chavin also mark the spread of
the Chavin phenomenon. This has precipitated a search for “fossil
index” markers, in the sense used by Adams (1965; Adams and Nissen,
1972) of the Chavin phenomenon— artifacts that one can identify as
pertaining exclusively to the spread of the Chavin cult of influence.
Theoretically, such “fossil indices” are supposed to be valuable chron-
ological tools, representing a relatively short-lived yet widespread entity
or a horizon (Willey, 1951:1 1; Willey and Phillips, 1958:29-34).

1983

Pozorski—Caballo Muerto Complex

23

Fig. 12. — Decorated sherds from Caballo Muerto; a-c are from Huaca Guavalito, d and
h are from Huaca Herederos Chica, e-f are from construction phase 2 of Huaca Curaca,
and g is from construction phase 1 of Huaca de los Reyes. Sherd a is combed decoration,
b has combing and applique bumps, c has applique bumps, d-e have incised raised
bands, /has zoned slashlike punctations, ^has an incised raised band as well as incision,
and h has applique bumps and punctation.

Unfortunately, theory, at present, does not correlate well with fact.
First, the Chavin phenomenon cannot be strictly called a horizon,
because estimates of its duration often range up to 1000 years (Engel,
1966:41; Lumbreras, 1970:41; Mason, 1969:43; Proulx, 1973:6; Rowe,

24

Annals of Carnegie Museum

VOL. 52

Fig. 13. — Decorated sherds from Caballo Muerto: a and / are from construction phase
2 of Huaca de los Reyes, b and d are from Huaca La Cruz, c is from construction phase
2 of Huaca Curaca, e is from Huaca Herederos Chica, g is from Huaca Guavalito, and
h is from construction phase 2 of Huaca Curaca. Neckless olla sherds a-b and ti-Z^have
incised rim or lip decoration (interiors of sherds are shown), c is a bottle spout rim with
incised decoration and g-h are decorated with dentate rocker sampling.

1967d[:21, 25; Sawyer, 1968:18-19; Willey, 1971:84-85), and even the
most conservative estimate would see it lasting for at least 400 years
(Lanning, 1974:76). Second, attempts to find chronological markers
other than actual Chavin style designs and motifs have been singularly
unsuccessful. This failure is most evident in ceramic decoration.

1983

PozoRSKi — Caballo Muerto Complex

25

Fig. 14.— Stirrup spout fragments from Caballo Muerto: a~g, i and k are from Huaca de
los Reyes construction phase 2, h is from Huaca La Cruz, and j is from Huaca de los
Reyes construction phase 1 .

Ceramic Evidence

Some investigators have designated certain ceramic motifs and modes
of decoration as Chavin. One of the most often cited set of Chavin
ceramic motifs is the circle, concentric circle, or circle and dot motif
(Burger, 1978:336; Fung, 1969: 1 38; Patterson, 1971:33; Proulx, 1973:
23-25; Rosas, 1976:571, 573; Shady, 1976:588; Willey and Corbett,
1954:37, 49), either as stamped or incised decorations. Another com-
monly cited Chavin motif is plain or dentate rocker-stamping (Collier,
1955:210; Flores, 1960:343; Fung, 1969:139; Matos, 1968:229; Pat-
terson, 1 97 1 :32-33). Less cited examples of Chavin ceramic decoration
include feline motifs (Matos, 1968:229; Patterson, 1971:33; Shady,
1975:588), pattern burnishing (Proulx, 1973:23; Rowe, 1960:628), and
combing (Patterson, 1971:32-33; Rosas, 1976:568, 572).

A review of published literature reveals that each of these “Chavin
markers” has been found in pre-Chavin or Initial Period contexts. For
example, incised circles, concentric circles, and circle and dot motifs
have been found in the highlands (Izumi and Sono, 1963:140, 153;
Lathrap, 1974:75; Rosas, 1976:568; Rosas and Shady, 1970:8) and on

26

Annals of Carnegie Museum

VOL. 52

Fig. 15. — Incised neckless olla rims from Huaca Herederos Chica.

the coast Patterson, 1968:423). Stamped circular decoration occurs in
Kotosh Kotosh levels at Kotosh (Izumi and Terada, 1972:186-188).
Dentate rocker-stamping is pre-Chavin at Chavin de Huantar (Burger,
1978:99-101) and Kotosh (Izumi and Terada, 1972:188-189), and
plain rocker-stamping occurs in Initial Period levels both in the high-
lands (Burger, 1978:101-102; Izumi, 1971:59,62, 69; Izumi and Sono,
1963:139, 155; Izumi and Terada, 1972:189, 307) and on the coast
(Patterson, 1971:32). In addition, feline motifs (Izumi, 1971:59, 69;
Kano, 1979:28-38; Rosas, 1976:568; Rosas and Shady, 1970:9; Shady,
1976:588), pattern burnishing (Rosas, 1976:568; Izumi and Terada,
1972:196-198), and combing (Patterson, 1971:32; Rosas, 1976:568;
Rosas and Shady, 1970:9) have all been found in pre-Chavin contexts.

All other decorative modes (except square rim castellation) found at
Caballo Muerto have been found in Initial Period levels at sites on the
coast and in the highlands. These modes include broad-line incision
(Burger, 1978:81, 1979:136-137; Fung, 1969:67-69; Izumi and Sono,
1963:139, 153-155; Izumi and Terada, 1972:188-193; Lumbreras,
1974a:51-52; Matos, 1968:228-229; Strong and Evans, 1952:289-291);
hne-line incision (Fung, 1969:71; Izumi and Terada, 1972:189-196;
Larco Hoyle, 1941:81-82; Strong and Evans, 1952:286-289); punc-
tation and zoned punctation (Burger, 1978:87-92; Grieder, 1975:105-
106; Izumi, 1971:62-69; Fanning, 1967:86; Reichlen and Reichlen,
1949:154; Rosas and Shady, 1970:9; Strong and Evans, 1952:283-284;
Willey, 1971:109); crosshatching (Fung, 1969:86; Patterson, 1971:32;
Strong and Evans, 1952:287-288); decorated raised ribs or bands

1983

PozoRSKi — Caballo Muerto Complex

27

Fig. 16. ““Sherds decorated with punctation {a-b, d) and zoned punctation (c) found at
Huaca Herederos Chica.

(Burger, 1978:94-96; Izumi and Terada, 1972: 192; Canning, 1967:86;
Rosas, 1976:568; Strong and Evans, 1952:277-282); modeling (Fung,
1969:67-69; Izumi and Sono, 1963:139, 154; Izumi and Terada, 1972:
194; Matos, 1966:515, 1968:229; Rosas and Shady, 1970:9); applique
bumps (Collier, 1962:413; Fanning, 1967:85; Shady, 1976:582); post-
fired paint (Burger, 1978:104; Fung, 1969:69; Izumi and Terada, 1972:
189, 192;Reichlen and Reichlen, 1949:154-155; Wallace, 1962:313);
rim lip incision (Shady, 1976:582); and graphite as a filler of incision
(Izumi and Terada, 1972:188, 193), in bands on zones (Engel, 1956:
103-104; Izumi and Sono, 1963:155; Fanning, 1967:85-86; Fum-
breras, 1974d!:54; Willey, 1971:109), and as an overall slip (Burger,
1978:103-104; Izumi and Sono, 1963:139, 155; Izumi and Terada,
1972:193).

Though virtually all of the decorative techniques present in the ce-
ramic sample from Caballo Muerto have been recovered from Initial
Period contexts in other parts of Peru, these data must be viewed in
conjunction with vessel forms and ware types in order to assess the
relationship Caballo Muerto had with other areas of Peru. Based on
all these characteristics, the ceramics associated with the Group I and
Group II mounds of Caballo Muerto, including Huaca de los Reyes,
appear to be most closely related to the Early and Middle Guanape
ceramics from Huaca Negra in Viru (Strong and Evans, 1952) and to
those from Fas Haldas south of Casma (Fung, 1969; Grieder, 1975),

28

Annals of Carnegie Museum

VOL. 52

Fig. 17. — Rim or lip incision on neckless olla sherds from Caballo Muerto. The interiors
of the sherds are shown: a is from Huaca Herederos Grande, b and d are from Huaca
La Cruz, and c and e are from construction phase 2 of Huaca de los Reyes.

two sites generally considered to be representative of the Initial Period
(Lumbreras, 1974^:51-52; Willey, 1971:109).

Ceramic data suggest that for true time markers of the Chavin phe-
nomenon one must rely on extant Chavin style designs as positive
proof That, however, can also be a mistake. The situation cannot be
better exemplified than at Caballo Muerto. At Huaca de los Reyes,
Chavin-related friezes are associated with numerous pottery decoration
techniques, most of which are present at earlier mounds. Use of graphite
in zones and within incision first appears in the Caballo Muerto com-
plex at Huaca de los Reyes. None of the pottery fragments, however,
including the graphite ones, exhibit definite Chavin style motifs. If the
friezes were not present at Huaca de los Reyes, there would be no
evidence that mound was Chavin-related since all of the associated
pottery decorative techniques are known from other parts of Peru in
pre-Chavin contexts.

Radiocarbon Evidence

The above dilemma points out the current chaotic state of the study
of early Andean ceramics. Definitions of just what is “Chavin” and
what is “pre-Chavin” seem to vary with each investigator. Absolute
dating techniques can help, and an examination of dates from early

1983

Pozorski—Caballo Muerto Complex

29

Fig. 18. —Punctated sherds from construction phase 2 of Huaca de los Reyes.

ceramic sites in the Andean area reveals enlightening patterns that are
particularly relevant to the dating of the Initial Period and the Early
Horizon.

Radiocarbon dates are available from four highland sites, Kotosh,
Chavin de Huantar, Shillacoto, and Pacopampa. At Kotosh, the Kotosh
Waira-Jirca pottery is dated by radiocarbon dates ranging from 1850
to 1050 B.C., with the investigators assigning a time range of 1500 to
1000 B.C. for the period (Izumi and Terada, 1972:308). The pre-Chavin
Kotosh Kotosh period pottery is associated with dates ranging from
1120 to 890 B.C., with an assigned time range of about 1000 to 800
B.C. (Izumi and Terada, 1972:309). For Kotosh Chavin pottery, dates
range from 1200 to 870 B.C., with an estimated time range of about
1000 to 300 B.C. (Izumi, 1971:59-62; Izumi and Sono, 1963:154-156;
Izumi and Terada, 1972:310).

At Chavin de Hauntar, early dates ranging from 1420 to 940 B.C.
have been reported (Amat, 1976:544; Lumbreras, 1970:133), but Bur-
ger (1981:596) feels that these dates may have been contaminated by
groundwater carbonates. More secure dating obtained by Burger (1 979:
154, 198 1 : 594-596) correlates well with the majority of dates obtained
by Lumbreras (1972:78). Burger (1981:596) dates the Initial Period or
Urabarriu occupation from 850 to 460 B.C. and the last two Early
Horizon phases, Chakinani and Janabarriu, as 460 to 390 B.C. and

30

Annals of Carnegie Museum

VOL. 52

390 to 200 B.C., respectively. The last phase, Janabarriu, is the one to
which Burger (1978:396, 1981:600) ascribes the spread of Chavin in-
fluence over a pan-Andean area.

The site of Shillacoto has Waira-jircalike pottery which has a radi-
ocarbon date of 1250 B.C. ± 80 (Izumi and Terada, 1972:308). Finally,
at Pacopampa, pre-Chavin pottery is associated with a date of 1835
B.C. ± 100 (Rosas, 1976:568).

On the coast, there is a strikingly different pattern. Relatively few
dates have been published for what has been called Cha /in or Early
Horizon related pottery, but extant dates range from 1090 to 342 B.C.
with the majority of the dates in the 800 to 700 B.C. range (Bird, 1951:
40; Collier, 1955:25, 1962:413; Fung, 1969:180-186; Ishida et al.,
1960:518; Kigoshi et al., 1961:92; Rowe, 1967^z:27~30). The more
abundant dates for what is defined as Initial Period pottery range from
1940 to 570 B.C., with most falling within the 1600 to 1100 B.C. time
span (Berger et al., 1965:347; Bird, 1951:40, Burger, 1981:594; Collier,
1955:25; Fung, 1969:180-186; Matos, 1968:230; Patterson, 1968:423;
Rowe, 1967a:26-30).

Hence, two distinct patterns emerge from the absolute dates available
for early ceramic sites in the Andean area. In general, the dates from
coastal sites are substantially older than ones from the highlands, even
when coastal and sierra sites that supposedly ceramically date to the
same time period are compared. There is strong evidence to indicate
that what is defined as the Initial Period or pre-Chavin pottery tradi-
tion, dominated by the neckless olla, goes back to at least 1800 B.C.
on the coast and lasts for 700 years or more. Sites in the highlands
assigned to the Initial Period are characterized by an additional variety
of vessel shapes and decoration distinct from the coast. The selva-
related Waira-jirca component at Kotosh and the early pottery of Pa-
copampa appear to be comparable in date to the coastal Initial Period
sites such as Caballo Muerto, Huaca Negra, Las Haldas, Ancon, Gar-
agay and La Florida. Other highland Initial Period components, how-
ever, at sites like Kotosh, Chavin de Huantar and Shillacoto date closer
to 1 000 B.C. or, as at Chavin de Huantar, 200 to 500 years later (Burger,
1981:596-599).

The same distinction between coast and highlands can be made for
Chavin-related sites attributed to the Early Horizon. Coastal sites date
closer to 1000 B.C., whereas sierra ones date nearer to 500 B.C. Clearly,
highland sites attributed to the Initial Period and the Early Horizon
date much later than coastal ones considered to be Initial Period and
Early Horizon.

With the above review, it is clear that, based on absolute dating,
only part of the Waira-jirca phase of Kotosh and the early phase of
Pacopampa can be assigned to the Initial Period according to Rowe’s

1983

PozoRSKi — Caballo Muerto Complex

31

(1960:627-631, 1962<2:49; Rowe and Menzel, 1967:vi-viii) definition
and sequence which places this chronologically between 2100 and 1400
B.C. No other sierra sites are sufficiently early to be correlated with
the master sequence of the lea Valley. Several coastal sites, on the other
hand, do date to the Initial Period. If, however, the dates of 1800 to
900 B.C. for the Initial Period according to Panning (1967:25) are used,
then some highland sites do belong in the Initial Period. Thus, the
problem becomes one of deciding where to place the dividing line
between the Initial Period and the Early Horizon. According to both
sequences, sites are present both on the coast and in the highlands that
date within the Early Horizon time span. However one looks at the
situation, though, the coastal sites consistently predate their highland
counterparts within the same general time period.

Reinterpreting the Chavin Phenomenon

In broader terms, how does one interpret this situation? What im-
portance can be placed on the Chavin phenomenon? Was it really a
unifying force? Present evidence suggests that it was not. In other words,
there has been too much emphasis on relating sites to the Chavin
phenomenon without objective analysis. Early civilization of the Pe-
ruvian coast was substantially different from that in the highlands, at
least as exemplified in ceramics and architecture. Coastal pottery was
dominated by the neckless olla tradition which lasted at least 1500
years from 1800 to 300 B.C. if not longer. Often this pottery is directly
associated with large corporate labor U-shaped mounds which in turn
are frequently associated with sunken circular forecourts. These ar-
chitectural patterns can be traced back in time to the Cotton Preceramic
period (Pozorski and Pozorski, 1979; Moseley, 1975:83; Williams, 1972,
1978-1980).

In the highlands, on the other hand, early pottery is generally later
in date, of a finer quality, and bears a great variety of vessel shapes
and decoration that are distinct from coastal pottery while showing
strong ties to the tropical forest (Lathrap, 1971:73-100). This is es-
pecially true at Kotosh (Izumi and Sono, 1963; Izumi and Terada,
1972) and at the recently excavated site of Huaricoto near Huaraz
(Burger and Burger, 1980). The one highland site that shows fairly
strong ceramic affinities with the coast is Pacopampa (Rosas and Shady,
1970:1-16, 1974:6-32) in the northern highlands.

In terms of monumental architecture, the highlands have a pattern
of small mounds and niched rooms that date as early as preceramic
times as seen at Kotosh (Izumi and Sono, 1963:40-70, 153-154; Izumi
and Terada, 1972:129-176), La Galgada in the upper Santa Valley
(Bueno and Grieder, 1979; Grieder and Bueno, 1 98 1 :45-48), and Huar-
icoto (Burger and Burger, 1980:27-28). Chavin de Huantar shows in-

32

Annals of Carnegie Museum

VOL. 52

Fig. 19. — Punctated and zone punctated sherds from construction phase 2 of Huaca de
los Reyes.

fluences both from the tropical forest and the coast in the form of
ceramics and iconography (Burger, 1978:393-402; Lathrap, 1971:73-
100). Given the late dates reported by Burger (1979:154-155, 1981:
592-602) which reinforce most of the dates obtained by Lumbreras
(1972:78), it is also clear that coastal influence at Chavin de Huantar
is reflected in the architecture— specifically the U-shaped mound form
and the sunken circular forecourt (Lumbreras, 1971:2, Fig. 2, 1974Z?:
39, 1977:2-14, Laminas I-III).

Interaction occurred between the highlands and the coast, but only
in a limited manner and only very late (that is, well into the first
millenium B.C.). Chavin de Huantar ceramics contain several coastal
elements, even in the earliest phases, and possible Cupisnique trade
pieces are present (Burger, 1978:274, 394). On the coast, however, few
clearly Chavin-associated highland ceramic traits have been reported,
at least at the most famous of the so called “Chavin” sites. Significantly,
for example, neither survey and excavation by Tello (1956:82) nor the
later work of Collier and Thompson (Thompson, 1964:207) turned up
any “Classic Chavin” pottery in the Casma Valley any nearer the coast
than Pallka, situated about 40 km inland. Casma is the valley that
contains Moxeke, noted for its “Chavin” adobe friezes, and Cerro
Sechin, as well as several other reputedly “Chavin” sites. In the neigh-
boring valley of Nepena, which contains Cerro Blanco and Punkuri,

1983

PozoRSKi— -Caballo Muerto Complex

33

Fig. 20. — Sherds from construction phase 2 of Huaca de los Reyes bearing zones and
bands of graphite paint.

both of which are said to have “Chavin” adobe friezes (Roe, 1974:37-
38), highland-related pottery is concentrated in the middle and upper
reaches of the valley (Daggett, personal communication; Proulx, 1973:
14-29), and no Early Horizon ceramic associations have been docu-
mented for the friezes of either Cerro Blanco or Punkuri (Tello, 1943).

However, when investigators speak of highland-coastal interaction
during the Early Horizon, the Chavin phenomenon on the coast is
invariably said to be represented by the iconography at Moxeke, Pun-
kuri, and Cerro Blanco, occasionally by the anomalous site of Cerro
Sechin, and by the Cupisnique pottery of Larco Hoyle from the north
coast— all of which are essentially undated and tied only by seriational
arguments to the sculptural sequence at Chavin de Huantar (Proulx,
1976:1-5; Rowe, 1962/?:13, 1967Z?:76; Roe, 1974:26-39, 1978:1-41).
Diseoveries at Huaca de los Reyes and Garagay near Lima have been
considered in this scenario by some authors who cite them as additional
coastal examples of Chavin cult influence. Roe (1978:7-8) has recently
stated that the Huaca de los Reyes friezes date to late phase D or early
phase EF of the Rowe sequence in spite of iconographic (Pozorski,
1975:211-251, 1976:63-92, 112-114, 1980:100-110) and absolute
dating evidence for Chavin de Huantar versus Huaca de los Reyes,
which make that correlation impossible (Burger, 1981:600; Pozorski,

34

Annals of Carnegie Museum

VOL. 52

c

Fig. 21. — Sherds from Huaca Guavalito decorated with graphite: a-b are painted with
bands of graphite, c-d each have an exterior graphite slip, and e has a red slip and graphite
paint within incised decoration.

1980:108-109). If one accepts Burger’s (1978:320, 1981:600) corre-
lation of his Janabarriu phase, the time of the spread of Chavin dom-
inance, with Rowe’s phase E and his suggested dates of 390 to 200
B.C., then the radiocarbon dates of about 1300 B.C. from Huaca de
los Reyes clearly do not match.

Although Burger (1981:598) accepts the Caballo Muerto dates, he
questions them because all were run on cana brava {Gynerium sagit- \
tatum) which might give anomalous results like sugar cane and other
grasses. However, Conrad (1974:739-748) had nine cana brava sam-
ples tested for ratios when dating compounds at Chan Chan

and found the C-13 average values to be —23.6 ± 0.5%, very close to
that for wood and considerably different from the — 1 2% value for sugar
cane and other grasses.

Rowe (1962Z?:5-6, 1967Z?:73-74) originally established his Chavin
seriation based on the master sequence of the lea Valley (Menzel et !
al., 1964). Absolute dates were later assigned to the sequence (Rowe

1983

PozoRSKi— Caballo Muerto Complex

35

Fig. 22.— Stamped circular decoration on sherds from Huaca Guavalito.

and Menzel, 1967:vi-viii), indicating that the Chavin style and the
mechanism that spread the style originated somewhere north of Ica in
the late Initial Period. The arrival of the Chavin style in the Ica Valley
signalled the start of the Early Horizon. The chart in Rowe and Menzel
(1967:vi-vii) indicates that phase AB, the earliest of the Chavin style
phases, begins in the Ancash area about 1500 B.C., in the late Initial
Period, and is not present until 1400 B.C. in Ica, the beginning of the
Early Horizon. This phase is then followed by phases C, D, and EE
spaced out over a period of several hundred years. Roe (1974) later
extended Rowe’s Chavin sedation to include most known Chavin style
objects and monuments in Peru. His extension of the original Rowe
seriation, however, is biased because 1) he assumes that all Chavin-
related elements emanate from Chavin de Huantar, and 2) he uncrit-
ically follows the original seriation, which was meant as a preliminary
suggestive guide (Rowe, 1967Z?:76), without accounting for associated
noniconographic evidence.

Nevertheless, Roe (1974:31, 1978:12-13) as well as Burger (1978:
389-391) have pointed out, at least indirectly, a major weakness of
Rowe’s original Chavin seriation by reinterpreting the entire Ica se-
quence as dating no earlier than phase D. This could very well be true
because the published radiocarbon dates for Early Horizon phases 3(459
B.C. ±214) and 9 (451 B.C. ± 1 10) of the Ocucaje sequence (Rowe,
1967(3:28-30) are within the range of Burger’s (1981:594-596) dating
of the Janabarriu phase at Chavin de Huantar which he correlates
iconographically with Rowe’s phase D (Burger, 1978:389-391). The
data from Chavin de Huantar and Ica suggest a possible relationship

36

Annals of Carnegie Museum

VOL. 52

between the two, but it is clear that neither contributed to the earlier
Initial Period developments on the north and central coasts. The pro-
posed absence of Chavin phases AB and C in the Ocucaje sequence of
lea, which was originally used to establish the Chavin sequence (Rowe,
1962^:5, 1967Z?:73), brings into question the validity of the Chavin
iconographic seriation.

The artifactual, architectural, iconographic, and radiocarbon evi-
dence from Huaca de los Reyes and Caballo Muerto in general are in
very good agreement with most of the evidence from coastal Peru for
Initial Period times. Looking at the coast as a whole, Caballo Muerto
is but one of several centers that arose in the first half of the second
millineum B.C. to control irrigation agriculture that was just beginning,
but rapidly spreading over most of the north and central coast of
(Moseley, 1974; Pozorski, 1976; Pozorski and Pozorski, 1979). Pop-
ulation centers shifted from the coastline, where subsistence in Cotton
Preceramic times had emphasized the gathering of marine resources,
inland to strategic points within the narrow coastal river valleys to
control canal intakes for irrigation agriculture. These corporate labor
mound sites are present in every valley on the north and central coast,
but a few centers probably became dominant over several valleys.
Contrary to Burger’s (1981:596-600) assessment, most of these sites
contain substantial occupational evidence and are not empty cere-
monial centers.

Little, if any, interaction occurred between the coast and highlands
from 1800 to 800 B.C., and the interaction accomplished after that
time appears to have been mostly one way, from the coast to the
highlands. Chavin de Huantar reflects coastal iconography in the form
of the human figure (Huaca de los Reyes, Moxeke, Cerro Sechin), the
feline (Huaca de los Reyes, Punkuri), and the Ofrendas monster (Gar-
agay). It also reflects coastal architecture in the form of the U-shaped
mound and the sunken circular forecourt. Coastal ceramic traits such
as stirrup spout vessels are also present at the site.

As Burger (1981:599-600) has demonstrated, when one speaks of
the Early Horizon and the spread of Chavin influence, he must realize
that the type site of Chavin de Huantar cannot be viewed as a purely
“Chavin” site, and and certainly not as the source for all related ar-
chitectural, iconographic, and ceramic elements. Instead, Chavin de
Huantar is a blending of elements from the coast, the highlands, and
the tropical forest and was apparently primarily a receiver, not a sender,
of what have come to be called “Chavin” traits. This interpretation is
quite fitting in view of its intermediate highland location between the
coast and the tropical forest (Burger, 1978:400; Silva, 1978:1-25).

Stylistic comparisons of iconography from diflerent regions must take
into account associated architectural and artifactual evidence as well

1983

Pozorski—Caballo Muerto Complex

37

as absolute dates. If one attempts to discuss the Chavin phenomenon,
then definitions and distinctions between the Initial Period and the
Early Horizon must be clarified, refined, and applied to all varieties of
archaeological evidence. Anything less would not take into account the
present state of knowledge. Conversely, if, as it is now beginning to
appear, such a tremendous unifying force or true “horizon” did not
exist during the early portion of the central Andean sequence, then
more profitable lines of inquiry should be pursued.

Acknowledgments

Fieldwork at Caballo Muerto was conducted under a permit given to the Chan Chan-
Moche Valley Project, directed by M. E. Moseley and C. J. Mackey, by the Peruvian
Institute Nacional de Cultura. I would like to thank the National Geographic Society
and the Institute of Latin American Studies at the University of Texas at Austin for
providing the funds that made the field research for this paper possible. My appreciation
goes to Genaro Barr who did the illustrations of the ceramics. This paper has also
benefitted from the comments and criticisms of Richard Burger.

Literature Cited

Adams, R. M. 1965. Land behind Bagdad: a history of settlement on the Diyala plains.
Univ. Chicago Press, Chicago, 187 pp.

Adams, R. M., and H. Nissen. 1972. The Uruk countryside: the natural setting of
urban societies. Univ. Chicago Press, Chicago, 241 pp.

AmatO., H. 1976. EstudiosarqueologicosenlacuencadelMosnay enel AltoMaranon.

41st Intemat. Congress Americanists, 3:532-544.

Berger, R., G. J. Ferguson, and W. F. Libby. 1965. UCLA radiocarbon dates IV.
Radiocarbon, 7:347.

Bird, J. B. 1951. South American radiocarbon dating. Pp. 37-49, in Radiocarbon
dating (F. Johnson, ed.), Mem. Soc. Amer. Arch., 8:1-65.

Bueno M., A., and T. Grieder. 1979. Arquitectura preceramica de la sierra norte.
Espacio, 1(5).

Burger, R. L. 1 978. The occupation of Chavin, Ancash, in the Initial Period and Early
Horizon. Unpublished Ph.D. dissert., Univ. California, Berkeley, 526 pp.

-. 1979. Resultados preliminarios de excavaciones en los distritos de Chavin de

Huantar y San Marcos, Peru. Pp. 133-155, in Arqueologia peruana: investigaciones
arqueologicas en el Peru, 1976 Seminario (R. Matos M., ed.), Lima, 213 pp.
. 1981. The radiocarbon evidence for the temporal priority of Chavin de Huan-
tar. Amer. Antiquity, 46:592-602.

Burger, R. L., and L. S. Burger. 1 980. Ritual and religion at Huaricoto. Archaeology,
33:26-32.

Carrion Cachot, R. 1948. La cultura Chavin, dos nuevas colonias: Kuntur Wasi y
Ancon. Rev. Mus. Nac. Antropologia y Arqueologia, 2:97-172.

Collier, D. 1955. Cultural chronology and change as reflected in the ceramics of the
Viru Valley, Peru. Fieldiana: Anthropology, 43:1-226.

. 1962. Archaeological investigations in the Casma Valley, Peru. 34th Intemat.

Congress Americanists, 1:411-417.

Conklin, W. J. 1 974. Pampa Gramalote textiles. Pp. 77-92, in Irene Emery roundtable
on museum textiles, 1974 proceedings: archaeological textiles (P. L. Fiske, ed.). The
Textile Museum, Washington, D.C., 210 pp.

Conrad, G. 1974. Burial platforms and related structures on the north coast of Peru:

38

Annals of Carnegie Museum

VOL. 52

some social and political implications. Unpublished Ph.D. dissert., Harvard Univ.,
Cambridge, 752 pp.

Engel, F. 1956. Curayacu, a Chavinoid site. Archaeology, 9:98-105.

. 1966. Geografia humana prehistorica y agricultura precolombina de la Que-

brada de Chilca, I. Universidad Agraria, Oficina de Promocion y Desarrollo, De-
partamento de Publicaciones, Lima, 1 10 pp.

Flores E., I. 1960. Wichqana, sitio temprano en Ayacucho. Pp. 335-344, in Antiguo
Peru, espacio y tiempo (R. Matos M., ed.), Libreria-Editorial Juan Mejia Baca,
Lima, 400 pp.

Grieder, T. 1975. A dated sequence of building and pottery at Las Haldas. Nawpa
Pacha, 13:99-112.

Grieder, T., and A. Bueno M. 1981. La Galgada: Peru before pottery. Archaeology,

34:44-51.

IsHiDA, E., ET AL. 1960. Andes, the report of the University of Tokyo scientific expe-
dition to the Andes in 1958. Bijitsu Shuppansha, Tokyo, 528 pp.

IzuMi, S. 1971. Development of the Formative culture in the Ceja de Montana of the
central Andes. Pp. 49-72, in Dumbarton Oaks Conference on Chavin, 1968 (E. P.
Benson, ed.), Dumbarton Oaks Research Library and Collection, Washington, D.C.,
124 pp.

IzuMi, S., AND T. Song. 1963. Andes 2: excavations at Kotosh, Peru, 1960. Kadokawa
Publishing Co., Tokyo, 210 pp.

IzuMi, S., AND K. Terada (eds.). 1972. Excavations at Kotosh, Peru. A report on the
third and fourth expeditions. Univ. Tokyo Press, Tokyo, 375 pp.

Kano, C. 1979. The origins of the Chavin culture. Studies in Pre-Columbian art and
archeology, 24:1-87.

Kigoshi, K., Y. Tomidura, and K. Endo. 1962. Gakushuin natural radiocarbon mea-
surements 1. Radiocarbon, 4:91-92.

Lanning, E. P. 1967. Peru before the Incas. Prentice-Hall, Englewood Cliffs, New
Jersey, 216 pp.

. 1974. Western South America. Pp. 65-86, in Prehispanic America (S. Gor-

enstein, ed.), St. Martin’s Press, New York, 165 pp.

Largo Hoyle, R. 1941. Los Cupisniques. Casa Editora La Cronica y Variedades,
Ltda., 259 pp.

Lathrap, D. 1971. The tropical forest and the cultural context of Chavin. Pp. 73-100,
in Dumbarton Oaks Conference on Chavin, 1968 (E. P. Benson, ed.), Dumbarton
Oaks Research Library and Collection, Washington, D.C., 124 pp.

Lumbreras, L. 1970. Los templos de Chavin, guia para el visitante. Publicacion del
Proyecto Chavin, Lima, 165 pp.

. 1971. Towards a re-evaluation of Chavin. Pp. 1-28, in Dumbarton Oaks

Conference on Chavin, 1968 (E. P. Benson, ed.), Dumbarton Oaks Research Library
and Collection, Washington, D.C., 124 pp.

. 1972. Los estudios sobre Chavin. Revista Mus. Nac., 38:73-92.

. 1974<2. The peoples and cultures of ancient Peru, translated by B. J. Meggers.

Smithsonian Institution Press, Washington, D.C., 248 pp.

— . 1974^. Informe de labores del proyecto Chavin. Arqueologicas, 15:37-55.

— : . 1977. Excavaciones en el templo antiguo de Chavin (sector R); informe de la

sexta campana. Nawpa Pacha, 15:1-38.

Mason, J. A. 1969. The ancient civilizations of Peru. Penguin Books, London, 335

pp.

Matos M., R. 1966. El Periodo Ceramico Inicial en la costa central del Peru. 36th
Intemat. Congress Americanists, 1:509-518.

. 1968. A Formative Period painted pottery complex at Ancon, Peru. Amer.

Antiquity, 33:226-232.

1983

PozoRSKi— Caballo Muerto Complex

39

Menzel, D., J. H. Rowe, and L. E. Dawson. 1964. The Paracas pottery of Ica, a study
in style and time. Univ. California PubL, American Archaeology and Ethnology,
50:1-382.

Moseley, M. E. 1 974. Organizational preadaptation to irrigation, the evolution of early
water-management systems in coastal Peru. Pp. 77-82, in Irrigation’s impact on
society (T. E. Downing and M. Gibson, eds.). Anthropological Papers of the Univ.
Arizona, 25, Tucson, 181 pp.

. 1 975. The maritime foundations of Andean civilization. Cummings Publishing

Co., Menlo Park, California, 131 pp.

Moseley, M. E., and L. Watanabe. 1974. The adobe sculpture of Huaca de los Reyes.
Archaeology, 27:154-161.

Patterson, T. C. 1968. Current research: highland South America. Amer. Antiquity,
33:422-424.

. 1971. Chavin: an interpretation of its spread and influence. Pp. 29-48, in

Dumbarton Oaks Conference on Chavin, 1968 (E. P. Benson, ed.), Dumbarton Oaks
Research Library and Collection, Washington, D.C., 124 pp.

PozoRSKi, T. 1975. El complejo de Caballo Muerto: los frisos de barro de la Huaca de
los Reyes. Revista Mus. Nac., 41:211-251.

. 1976. Caballo Muerto: a complex of early ceramic sites in the Moche Valley,

Peru. Unpublished Ph.D. dissert., Univ. Texas, Austin, 473 pp.

. 1980. The Early Horizon site of Huaca de los Reyes: societal implications.

Amer. Antiquity, 45:100-1 10.

. 1982. Early social stratification and subsistence systems: the Caballo Muerto

complex. Pp. 225-253, in Chan Chan: Andean desert city (M. E. Moseley and K.
C. Day, eds.). School of American Research, Advanced Seminar Series, Univ. New
Mexico Press, Albuquerque, 373 pp.

PozoRSKi, S., AND T. PozoRSKi. 1979. An early subsistence exchange system in the
Moche Valley, Peru. J. Field Archaeology, 6:413-432.

Proulx, D. 1973. Archaeological investigations in the Nepena Valley, Peru. Univ.
Massachusetts, Dept. Anthropology Research Report, 13:1-293.

. 1976. The Early Horizon of north coastal Peru: a review of recent develop-
ments. El Dorado, 1:1-14.

Reichlen, H., and P. Reichlen. 1949. Recherches archeologiques dans les Andes de
Cajamarca. J. Soc. Americanistes, 38:137-174.

Roe, P. 1974. A further exploration of the Rowe Chavin sedation and its implications
for north central coast chronology. Dumbarton Oaks Studies in Pre-Columbian Art
and Archaeology, 13:1-80.

. 1978. Recent discoveries in Chavin art: some speculations on methodology

and significance in the analysis of a figural style. El Dorado, 3:1-41.

Rosas, H. 1976. Investigaciones arqueologicas en la cuenca del Chotano. 4 1 st Intemat.
Congress Americanists, 3:137-174.

Rosas, H., and R. Shady S. 1970. Pacopampa: un complejo temprano del Periodo
Formative peraano. Arqueologia y Sociedad, 3:1-16.

. 1974. Sobre el Periodo Formative en la sierra del extreme norte del Peru.

Arqueologicas, 1 5 : 6-3 5 .

Rowe, J. H. 1960. Cultural unity and diversification in Peruvian archaeology. Pp. 627-
631, in Men and cultures, selected papers, 5th International Congress of Anthro-
pological and Ethnological Sciences (A. F. C. Wallace, ed.), Univ. Pennsylvania
Press, Philadelphia, 810 pp.

. 1962a. Stages and periods in archaeological interpretation. Southwestern J.

Anthropology, 18:40-54.

. 1962Z). Chavin art, an inquiry into its form and meaning. The Museum of

Primitive Art, New York, 23 pp.

40

Annals of Carnegie Museum

VOL. 52

. 1967<2. An interpretation of radiocarbon measurements on archaeological

samples from Peru. Pp. 16-30, in Peruvian archaeology, selected readings (J. H.
Rowe, and D. Menzel, eds.). Peek Publications, Palo Alto, 320 pp.

. 1961 b. Form and meaning in Chavin art. Pp. 72-103, />z Peruvian archaeology,

selected readings (J. H. Rowe, and D. Menzel, eds.). Peek Publications, Palo Alto,
320 pp.

Rowe, J. H., and D. Menzel (eds.). 1967. Peruvian archaeology, selected readings.
Peek Publications, Palo Alto, 320 pp.

Sawyer, A. 1968. Mastercraftsmen of ancient Peru. Solomon R. Guggenheim Foun-
dation, New York, 1 15 pp.

Shady, S., R. 1976. Investigaciones arqueologicas en la cuenca del Utcubamba. 41st
Intemat. Congress Americanists, 3:579-589.

Silva S., J. E. 1978. Chavin de Huantar: un complejo multifunctional. Universidad
Nacional Mayor de San Marcos Gabinete de Arqueologia, Colegio Real Serie In-
vestigaciones, 1:1-31.

Strong, W. D., and C. Evans. 1952. Cultural stratigraphy in the Viru Valley, northern
Peru. Columbia University Studies in Archaeology and Ethnology, 4:1-373.
Tello, j. C. 1943. Discovery of the Chavin culture in Peru. Amer. Antiquity, 9:135-
160.

. 1956. Arqueologia del Valle de Casma, culturas Chavin, Santa o Huaylas

Yunga y Sub-Chimu. Univ. San Marcos, Lima, 344 pp.

. 1 960. Chavin, cultura matriz.de la civilizacion andina, primera parte, 2. Univ.

San Marcos, Lima, 425 pp.

Thompson, D. 1964. Formative Period architecture in the Casma Valley, Peru. 35th
Intemat. Congress Americanists, 1:205-212.

Wallace, D. T. 1962. Cerrillos, an early Paracas site in Ica, Pern. Amer. Antiquity,
27:303-314.

Willey, G. R. 1951. The Chavin problem: a review and critique. Southwestern J.
Anthropology, 7:103-144.

. 1971. An introduction to American archaeology, volume two. South America.

Prentice-Hall, New Jersey, 559 pp.

Willey, G. R., and J. R. Corbett. 1954. Early Ancon and early Supe culture: Chavin
horizon sites on the central Pemvian coast. Columbia Univ. Studies in Archaeology
and Ethnology, 3:1-178.

Willey, G. R., and P. Phillips. 1958. Method and theory in American archaeology.
Univ. Chicago Press, Chicago, 270 pp.

Williams L., C. 1972. La difusion de los pozos ceremoniales en la costa peruana.
Apuntes, 2:1-9.

. 1978-1980. Complejos de piramides con planta en U, patron arquitectonico

de la costa central. Revista Mus. Nac., 44:95-1 10.

7. 73

ISSN 0097-4463

ANNALS

()/■ CARNEGIE MUSEUM

CARNEGIE MUSEUM OF NATURAL HISTORY

4400 FORBES AVENUE ® PITTSBURGH, PENNSYLVANIA 15213

VOLUME 52 30 MARCH 1983 ARTICLE 2

STUDIES ON NEARCTIC EUCHLOE. PART 8.
DISTRIBUTION, ECOLOGY, AND VARIATION OF
EUCHLOE OLYMPIA (PIERIDAE) POPULATIONS

Harry K. Clench’

Associate Curator, Section of Insects

Paul A. Opler^

Abstract

The biology, ecology, and geographic variation of the Olympia marble {Euchloe olym-
pia), a pierid butterfly, are described and discussed. This species occurs in relatively open
native situations with well-drained rocky or sandy substrates where its caterpillar hosts,
rock cresses {Arabis: Brassicaceae), may be found. The butterfly ranges from Maryland
east to Montana and from southern Ontario south to central Texas. Several groups of
populations or isolates each have distinctive traits relating to their caterpillar host plants,
habitat types and adult phenotypes. A statistical analysis of several adult wing characters
is presented, and the results are related to possible alternative taxonomic treatments.
The subspecific name rosa is best treated as a synonym of the nomenotypic species.

Introduction

Euchloe olympia Edwards is a distinctive, single-brooded pierid with
a range covering about half of the conterminous United States. Within
this range it is usually a highly localized insect with particular habitat
requirements.

Although the most recent revisionary treatment considers E. olympia
undeserving of infraspecific designation (Opler, 1966), other authors

‘ Deceased. Send reprint requests to second author.

^ Address: Division of Biological Services, U.S. Fish and Wildlife Service, Department
of the Interior, Washington, D.C. 20240.

Submitted 16 September 1982. T -Y tTi ■■

41

r-

■ ; o

Z/ftRftPlF.S

42

Annals of Carnegie Museum

VOL. 52

persist in suggesting such names may be merited (Howe, 1974; Wagner,
1977). Nevertheless, there does exist a certain amount of variation
within and between populations that is best treated by referring to ag-
gregates of populations or “isolates.”

In this paper we present and discuss the distribution of Euchloe
olympia, considering several biogeographic isolates. We then consider
the ecological attributes and variation shown by these groups of pop-
ulations.

Distribution

The total distribution of Euchloe olympia extends from southern
Manitoba and southern Ontario south to central Texas and from central
Montana, Wyoming, and Colorado east to West Virginia, Virginia,
Maryland, and Pennsylvania (Fig. 1).

It can be seen that there are several distinctive clusters or groupings
of populations which we will refer to hereinafter as “isolates” (Fig. 1);
these are (1) Appalachian isolate, (2) Great Lakes “dune” and “inland”
isolates (Wagner, 1977), (3) St. Louis isolate, (4) Texas isolate, (5)
Arkansas River isolate, (6) Missouri River isolate, and (7) Front Range
isolate.

Appalachian isolate.— series of E. olympia populations extends
from extreme southern Pennsylvania south through the mountains of
western Maryland, northwestern Virginia, and eastern West Virginia.
Colonies in central West Virginia, including Kanawha— olympiads
type locality— are best included with this isolate, although they may
be remnants of a once more extensive series of Ohio River drainage
populations.

Great Lakes isolates.— series of populations, which includes
colonies of diminutive individuals along lakeshore dunes, inland col-
onies of larger individuals, and those that appear intermediate, has
been described in detail by Wagner (1977). Populations occur in On-
tario, Michigan, northern Indiana, northern Illinois, northeastern Iowa,
Wisconsin, and Minnesota.

St. Louis isolate. the confluence of the Mississippi and Mis-

souri Rivers occurs a tight constellation of river bluff colonies (Mieners,
1939, 1957). !

Texas isolate. — Included here are several Texas populations which !|
occur near the Sabine, Trinity, Brazos, and Colorado rivers, all of which
flow into the Gulf of Mexico.

Arkansas River isolate.— This series of populations includes those |:
found along the Arkansas River and its tributaries, including the Ca- [
nadian River in northern Texas, as well as those found near the Ouachi-
ta and White rivers in Arkansas. Populations near the Arkansas in
central Colorado are placed with the Front Range isolate.

1983

Clench and Opler — Studies of Euchloe

43

i

Fig. 1. — Distribution of Euchloe olympia.

44

Annals of Carnegie Museum

VOL. 52

Missouri River isolate.— AW populations along the Missouri River,
and its tributaries, except those along the Upper Platte River drainage
in central Colorado and Wyoming, are included with this isolate. Also
put here is a population found along the Des Moines River in central
Iowa.

Front Range — Populations found in the mountains at the

western edge of the plains in central Colorado and Wyoming constitute
this isolate.

Ecology

Euchloe olympia is single brooded throughout its range, flying for a
brief period in early spring. Wherever it occurs it is among the first
emerging butterflies of the year, along with Incisalia, Anthocharis mi-
dea, the first Celastrina, the first Pieris rapae, and the first Erynnis. At
any single locality the average flight period seems to be no more than
20 to 25 days, unusually brief for a butterfly.

The onset of emergence is retarded, on average, about 4.5 days per
degree of latitude as one proceeds northward (Fig. 2). At any given
latitude, however, the first emergence may vary as much as 20 days or
so, depending on elevation, proximity to large water bodies, slope
exposure, year to year climatic variation, and other factors. During a
typical year olympia is on the wing, somewhere, from probably late
February, at the southern limit of its range, to early July, at its northern
limit or in the Rockies. The latest date of capture known to us is 18
July in Boulder County, Colorado, elevation unknown (Brown et al.,
1956).

With one possible exception, all known larval hosts are Arabis (Bras-
sicaceae), a primarily western genus. Each isolate seems to differ in
habitat type, flight habits, larval food plant, and adult nectar source.
As far as these attributes are known for each geographic isolate they
are discussed below.

Appalachian isolate. — A unique environment known as the Appa-
lachian shale barrens extends discontinuously from south-central Penn-
sylvania southwesterly to southern West Virginia and southwestern
Virginia (Platt, 1951). This ecological formation, characterized by a
number of endemic plant species of western affinities, is closely asso-
ciated with outcrops of the Devonian Braillier shale formation. A num-
ber of Appalachian butterflies occur in close association with these
shale barrens, and, of course, Euchloe olympia is among them. The
others are Erynnis brizo, Polites mystic, Hesperia sassacus, Hesperia
metea, Pyrgus centaureae, Calephelis borealis, Fixsenia Ontario, Glau-
copsyche lygdamus, and Phyciodes pascoensis.

In the barrens, E. olympia occurs in or near open, low pine or pine-
oak forests on south or southwest facing rocky shale slopes. It often

1983

Clench and Opler— Studies of Euchloe

45

Fig. 2. — Seasonal occurrence of adult Euchloe olympia as a function of latitude dem-
onstrating delayed emergence northward.

ventures to the wood edges and is frequently found along traversing
roads and trails.

On 27 May 1979, larvae were found by Opler feeding on Arabis
serotina Steele, a shale barrens endemic, near Sugar Grove, Pendleton
County, West Virginia. Other species of Arabis, including A. lyrata L.,
occur in shale barrens localities and may be utilized. A record of Euch-
loe olympia larvae feeding on Sisymbrium sp. at Coalburgh, West
Virginia (Edwards, 1891) is dubious. The host might have been Arabis
glabra. Clench found adults to nectar at Phlox subulata.

Wagner (personal communication) found a population of Euchloe
\ olympia residing on a limestone pavement habitat in the White Shoals
! District, Lee County, southwestern Virginia. Here, where the butterflies
j have been found from 25 April to 9 May, the habitat is characterized
i by stunted trees growing on limestone pavement with scattered boul-
ders. Red cedar (Juniperus virginiana) is especially prominent. The
' larval host at this locality is Arabis laevigata. It is possible that colonies
I in central West Virginia and Kentucky may be found in similar situ-
I ations rather than on typical shale barrens.

j Great Lakes dune isolate. — Known only from clusters of colonies in
i dunes at the upper edge of Lake Huron in Ontario and along the
I southern end of Lake Michigan in Wisconsin, Illinois, Indiana, and

46

Annals of Carnegie Museum

VOL. 52

Michigan is the diminutive dune form of E. olympia discussed in detail
by Shull (1907) and Wagner (1977).

Shull (1907) describes in great detail the adult flight habits of E.
olympia at the Indiana dunes. He states that “it flies near the ground,
following a rather uncertain course, dancing tremulously (sic!) this way
and that.” He also describes how E. olympia flies into the wind, par-
ticularly on the windy days which are so frequent along the lakeshore.
Shull found eggs on Arabis lyrata at the Indiana dunes and raised larvae
to pupation. He described the life history in detail (see also Opler,
1974). In the field, one larva was observed to pupate on the stalks of
Andropogon scoparius (Michx.) (Poaceae). Nielsen raised larvae found
on Arabis lyrata at a dune locality in Berrien County, Michigan (Opler,
1974) and E. olympia was found associated with this same plant on
dunes at Great Bend, Ontario (Wagner, 1977).

At the Indiana dunes location, Shull found adults to alight almost
exclusively on the flower clusters of Arabis lyrata, which is presumed
to be the principal nectar source there. M. C. Nielsen (personal com-
munication) found E. olympia in Berrien County, Michigan to fly along
the lakeward face of dunes 30 to 50 ft inland from the beach. Here, as
at Shull’s Indiana locality, E. olympia nectared at Arabis lyrata.

Great Lakes inland — Included here are a mixed lot of pop-

ulations with quite differing ecological features. Populations at inland
stations in Michigan, at least those in Allegan, Montcalm, and Ros-
common counties, are restricted to prairies, whereas in Ontario at least
one population in Garden Township is located on open limestone
barrens (Wagner, 1977). Nielsen (personal communication) reports that
at several sites in the northern portion of Michigan’s lower peninsula
E. olympia is usually found in openings within scrub oak-jack pine
habitat. At one site in Otsego County, he found E. olympia occurring
in an old pine burn, which had been colonized by mosses, lichens,
Andropogon, and Arabis.

In Minnesota, Masters (personal communication) found E. olympia
in deciduous woods in Pine and Chisago counties, while reporting it
from open hilltops, surrounded by deciduous woods, overlooking the
Mississippi River in Goodhue, Houston, and Winona counties, Min-
nesota. *

The usual host for these inland colonies is Arabis drummondii Gray.
Larvae have been found on this plant in Allegan, Montcalm, and Ros-
common counties, Michigan (Opler, 1974; Wagner, 1977), while near |
Killaloe, Ontario, Euchole olympia was observed ovipositing on Arabis i
glabra by Lindsay (1971). I

St. Louis isolate.— colonies are found associated with low,
sparse, open forest on the cherty summits of long ridges. There, adults '
fly low and seem confined to ridgetops (Meiners, 1937, 1939, 1956; |

1983

Clench and Opler— Studies of Euchloe

47

Table \. — Three characteristics of populations o/Euchloe olympia.

Forewing length

(cm)

Forewing apex

Forewing bar

Isolate

Sex

N

Mean

SD

Mean

SD

Mean

SD

Appalachian

<3

18

1.81

0.18

4.4

1.46

0.11

0.02

9

20

1.87

0.13

4.9

1.39

0.13

0.02

Michigan (inland)

<3

5

1.84

0.19

4.6

1.14

0.11

0.03

9

1

2.06

~

4.0

__

0.15

—

Great Lakes dunes

<3

13

1.72

0.12

4.9

0.95

0.09

0.02

9

5

1.70

0.06

5.2

0.45

0.11

0.01

Wisconsin

<3

7

1.74

0.07

3.0

1.00

0.08

0.01

9

5

1.80

0.12

3.2

1.30

0.10

0.02

Colorado Front Range

<3

26

1.79

0.08

2.7

0.89

0.10

0.02

9

10

1.84

0.12

3.1

1.20

0.13

0.03

Missouri River

<3

31

1.86

0.12

2.5

0.77

0.11

0.02

9

8

1.96

0.11

2.8

0.71

0.12

0.02

St. Louis

<3

7

1.96

0.07

4.4

0.98

0.12

0.02

9

3

1.91

0.14

4.3

0.58

0.12

0.01

Arkansas River

5

17

1.89

0.10

2.8

1.30

0.12

0.02

9

4

2.03

0.16

3.0

1.15

0.14

0.02

Texas

5

2

1.93

0.08

2.0

0.00

0.13

0.01

Total

<3

124

1.83

3.3

—

0.11

—

9

58

1.87

~

3.9

-

0.13

-

Arnhold, 1953), where its larval food plant Arabis viridis occurs (Mei-
ners, 1939).

Texas isolate. — IAwXq has been reported concerning this isolate ex-
cept that it is associated with limestone outcroppings near Dallas (Free-
man, 1959).

Arkansas River isolate.— The only ecological data concerning pop-
ulations that constitute this isolate is that adults are found in bottom-
land deciduous forest. Such observations have been made in Alfalfa,
Payne and Woodward counties, Oklahoma (McCoy, personal com-
munication), as well as in Newton County, Arkansas (Masters, personal
communication).

Missouri River isolate.— isolate covers the largest geographic
expanse, and the ecological characteristics of this isolate may be just
as wide-ranging. Masters (personal communication) found E. olympia
on a hilltop in jack pine forest at the Riding Mountains, Manitoba.

Front Range isolate. — Toc?i\ populations of Euchloe olympia in the
vicinity of the Colorado Front Range are found on open ridges or low
hills in short grass prairie habitat, often not too distant from cotton-
wood-lined streams (F. M. Brown, personal communication).

48

Annals of Carnegie Museum

VOL. 52

Table 2.— Differences among population o/Euchloe olympia in mean forewing length.
Student’s t (degrees freedom), P < 0.05*, <0.01**, <0.001.***

Sex

Isolate

Great Lakes
dunes

Colorado
Front Range

Missouri

River

Arkansas

River

Males

Appalachians

1.61 (29)

1.00 (42)

-1.10(47)

1.66 (35)

Great Lakes dunes

-2.33 (37)*

-3.34 (42)**

-4.32 (28)***

Colorado Front Range

-2.27 (55)*

-3.42 (41)**

Missouri River

-1.02 (46)

Females

Appalachians

0.59 (28)

-1.84 (26)

Colorado Front Range

-2.33(16)*

Variation

It is our intent here to analyze three physical characters and their
variation between isolates. On this basis, we may comment on the
prudence of recognizing any subspecific taxa. The three characters (Ta-
ble 1) measured were (1) forewing length (base to apex), (2) width of
forewing discal black bar, and (3) degree of black infuscation on fore-
wing apex above— judged on a relative scale of 1 to 7 with exemplars
as standards as follows: (a) apical area almost unmarked white: a minute
black subapical costal spot only; (b) costal spot, subapical (faint) and
spot at end of M3 (faint), apex with at most only a few black scales; (c)
costal and M3 spots strong, slight apical infuscation; (d) costal and M3
spots strong and with some connecting fuscous between them, apex
distinctly infuscate; (e) costal and M3 spots strong and connected; apical
infuscation strong; (f) apical and subapical infuscations strong, slightly
fused, (g) apical area almost solidly infuscate: a minute white patch on
costa only.

The results of this analysis are shown on Table 1, whereas Stu-
dent’s-! analysis was carried out between paired mean values for isolates
with adequate sample sizes (Tables 2, 3, 4).

The sexes were treated separately in these analyses, because females
were significantly larger (t = 2.15, P < 0.05), had significantly broader
discal bars (t = 5.15, P ^ 0.001), and had significantly darker apical
infuscation (t = 2.58, P < 0.05).

Only the Missouri River-Arkansas River pair did not show signifi-
cant differences in any character, whereas the Appalachian-Great Lakes
dune pair differed significantly only in forewing discal bar width and
the Appalachian-Front Range pairing differed significantly only in the
degree of apical infuscation. All other pairings differed significantly in
at least two of the three characters. The most dissimilar comparisons
were between the Great Lakes dune isolate and either the Missouri
River or Arkansas River isolates, there being significant differences in
the central tendency of all three characters.

1983

Clench and Opler— Studies of Euchloe

49

Table 3.— Differences among populations o/ Euchloe olympia in mean forewing apex.
Student’s t (degrees freedom), P <0.05*, <0.01**, <0.001.***

Sex

Isolate

Great Lakes
dunes

Colorado
Front Range

Missouri

River

Arkansas

River

Males

Appalachians

1.15(29)

4.89 (42)***

6.00 (47)***

3.47 (35)**

Great Lakes dunes

7.32 (37)***

8.15 (42)***

5.03 (28)***

Colorado Front Range

20.77 (55)

-0.33(41)

Missouri River

-0.94 (46)

Females Appalachians

3.40 (28)**

4.04 (26)***

Colorado Front Range

0.73 (16)

Size, which may be at least partially determined by environment,
was smallest in more montane, northern, or lakefront isolates, whereas
low elevation southern populations had larger individuals.

Degree of apical infuscation, a character asserted to separate sub-
species rosa, was generally greatest in the eastern-most populations and
least in western and southern isolates, although the St. Louis grouping
is unusual in having such dark infuscation due to its proximity to the
eight prairie isolates. Additionally, the Michigan inland prairie pop-
ulations, which are similar to the Missouri River and Arkansas River
isolates in facies (Wagner, 1977), occur in close proximity to the small,
dark dunes isolate.

Discussion

As we have seen with the possible exception of the similar Missouri
River and Arkansas River isolates, each of Euchloe Olympia’s isolates
occurs in a unique ecological setting and has its constituent individuals
displaying somewhat different phenotypic traits.

As to application of subspecific epithets, one is left with three choices.
One could apply a separate subspecies name to each isolate, as might
have been done in former times. Alternatively, one might restrict no-

Table ^.—Differences among populations o/ Euchloe olympia in mean forewing bar.
Student’s t (degrees freedom), P <0.05*, <0.01**, <0.001***.

Sex

Isolate

Great Lakes
dunes

Colorado

Front Range

Missouri

River

Arkansas

River

Males

Appalachians

2.56 (29)*

1.91 (42)

-0.27 (47)

-1.48 (35)

Great Lakes dunes

1.05 (37)

-3.56 (42)***

-3.87 (28)***

Colorado Front Range

-2.86 (55)**

-3.37(41)**

Missouri River

-1.23 (46)

Females

Appalachians

0.39 (28)

1.86 (26)

Colorado Front Rnage

0.86(16)

50

Annals of Carnegie Museum

VOL. 52

menotypic olympia to isolates with dark apical infuscation, that is,
Applachian, Great Lakes dunes, and St. Louis isolates, while applying
the subspecific name rosa to all lightly marked isolates. Finally, one
might choose to apply no subspecies names whatsoever. We have cho-
sen the latter course, because we feel the first option would be absurd
and the second would group together heterogeneous, polytopic sets of
populations which have responded to local environments and may not
be each other’s closest relatives. Of course, our unofficial isolate ter-
minology may be used in an informal sense to recognize these locally
adapted ecophenotypes. We feel that our discourse on the ecological
and phenetic variation of E. olympia to be of more critical importance
than the description of six or seven weakly differentiated subspecies.

Acknowledgments

We are grateful to the following for their valuable and varied assistance. Some provided
specimens, others gave us records, still others information on habits, occurrence and
other aspects of the species.

F. Martin Brown (Colorado Springs), Patrick J. Conway (Downer’s Grove, Illinois),
Charles V. Covell, Jr. (Univ. Louisville, Kentucky), Christopher Durden (Austin, Texas),
Ronald L. Huber (St. Paul, Minnesota), Roy O. Kendall (San Antonio, Texas), John H.
Masters (Valencia, California), Clarence J. McCoy, Jr. (Carnegie Museum of Natural
History), Burt L. Monroe, Jr. (Univ. Louisville, Kentucky), Mogens C. Nielsen (Lansing,
Michigan), the late William E. Sicker (Madison, Wisconsin), R. E. Stanford (Denver,
Colorado), and Robert W. Surdick (Pittsburgh, Pennsylvania).

We also thank Frederick H. Rindge (American Museum of Natural History) and
William D. Field (Smithsonian Institution) for allowing us to use material.

Locality Data

Abbreviations -- AMNH, American Museum of Natural History, New York; CAS,

California Academy of Sciences, San Francisco; CJM, Clarance J. McCoy, Jr.; CM,
Carnegie Museum of Natural History; CNC, Canadian National Collection, Ottawa; ex,
from, or from the collection of; JHM, John H. Masters; leg., collected by; MCN, Mogens
C. Nielsen; MNH, Natural History Museum, Washington, D.C.; NWS, Northwestern
State College, Alva, Oklahoma; OAS, Oakley A. Shields; OSU, Oklahoma State Uni-
versity, Stillwater; ROK, Roy O. Kendall; SS, Season Summary (News of the Lepidop-
terists’ Society, various years).

Canada

Alberta. Onefour 3. vi.l956 {leg. E. E. Stems, CNC).

Manitoba. Riding Mountain [town] 24.vi.1967 {leg. J. H. Masters, JHM); Brandon
(Hanham, 1 900); McCreary (per JHM).

Ontario. Killaloe [ca. 25 mi SW Pembroke] (SS); “La Cloche” [prob. Grande Cloche
Id., Georgian Bay] 28.V.1937 {leg. ?, CNC); Strawberry Id., Georgian Bay lO.v.1942 {leg.
N. J. Durand, CNC).

Saskatchewan. Regina 1885 {leg. “N.H.G.,” CNC).

United States

Arkansas. Carroll Co.: nr. Eureka Springs 18.iv.l965 {leg. J. R. Heitzman, PAO).
Garland Co.: Hot Springs 5.iv. 1932 {leg. L. I. Hewes, CAS). Madison Co.: Withrow

1983

Clench and Opler— Studies of Euchloe

51

Springs St. Park (SS). Newton Co.: Lost Valley, nr. Ponce iv.l965 {leg. J. H. Masters,
JHM). Washington Co.: Fayetteville {leg. L. Paulissen, per iHM).

Colorado. Arapahoe Co.: Highline Canal 29.iv.1961 {leg. R. Pyle, PAO). Boulder
Co..' (Brown etal. 1956); Lefthand Canyon 30.V.1957 {leg. D. Eff, PAO); Sunshine Canyon
4. V. 1962 {leg. D. Eff, OAS); Boulder 19.V.1936 {leg. T. N. Freeman, CNC); Boulder,
5500 ft, 7.vi.l961 {leg. M. R. Mackay, CNC); 4 mi NW Boulder, 6900 ft, 8.vi.l961 {leg.
J. R. Stainer, CNC). Denver Co.: Denver (nfd, CNC). Douglas Co.: (Brown et al., 1956).
El Paso Co.: (Brown et al., 1956). Eremont Co.: (Brown et al., 1956). Jefferson Co.:
(Brown et al., 1956); Clear Creek Canyon, W. Golden lO.v.1959 {leg. D. Eff, OAS); nfd.
8000 ft, 28.V.1961 (leg. “D.A.J.,” OAS); Lookout Mountain 17.V.1936 {leg. L. 1. Hewes,
CAS). Prowers Co.: (Brown et al., 1956). Note: In most instances. Brown et al. (1956)
give specific localities under the respective counties. These are not repeated here.

Illinois. Bureau Co. (nfd, per P. J. Conway). Cook Co.: Chicago 3. v. 1908 {leg. A.
Kwiat, CM). Lake Co.: northeastern part, 17.V.1914, 17.V.1919, 19.V.1923 (all leg. H.
M. Bower, CM, CNC); Zion {legl, CNC).

Indiana. Lake Co.: Gary 3.V.1908, 10. v. 1910 (both leg. A. Kwiat, CM); Whiting {leg.
?, ex W. H. Edwards, CM). Porter Co.: between Clarke Jet. and Pine (Shull, 1907). Not
located: Miller (s) (“sand dunes,” Leussler, 1938), also 6.v. 1 923 {leg. G. M. Dodge, CAS);
“N. Ind.” {leg. ?, ex W. H. Edwards, CM [one of these specimens figured, Holland (1931)
pi. 67 fig. 28]).

Iowa. Eremont Co.: Waubonsie St. Park 10. v. 1958 {leg. J. A. Ebner, ex L. D. Miller,
CM). Polk Co.: Des Moines (SS).

Kansas. Cowley Co.: nfd, 18,20.iv.l941, 12,17.iv. 1942 {leg. Stallings and Turner,
CAS; also, ex ANSP, CM). Douglas Co. (Field, 1938). Greenwood Co. (Field, 1938).
Harper Co. (Field 1938). Johnson Co.: Camp Towanyak, nr. Tulle Creek, late iii-early
iv {leg. J. H. Masters, JHM); Shawnee Mission Park 23.iv.1962 {leg. J. R. Heitzman,
OAS), no date {leg, J. H. Masters, JHM): S of Prairie Village 22. iv. 1965 {leg J. R.
Heitzman, PAO); Shawnee 23.iv.1964 {leg. J. H. Masters, PAO). Leavenworth Co. (Field,
1938). Pottawotamie Co. (Field, 1938). Reno Co. (Field 1938^ Scott Co. (Field, 1938).
Shawnee Co.: Topeka 20, 26.iv.4,5.v. 1941 {leg. R. H. Whittaker, CNC). Sumner Co.:
nfd, 14.iv {leg. ?, CM); also, nfd (Field, 1938). Wynadotte Co.: Quivera Spring, late iii-
early iv {leg. J. H. Masters, JHM).

Kentucky. Eloyd Co.: Jenny Wiley State Park 26.iv.1969 (C. V. Coveil, Jr. and B. L.
Monroe). Green Co.: Crailhope (C. Cook, 1948).

Maryland. Allegany Co.: nr. Flintstone, 12.V.1955 (Simmons, 1957); Green Ridge
State Forest, ca. 6 mi E Flintstone 9,10.v.l964 {leg. L. D. Miller and H. K. Clench, CM),
11.V.1965 {leg. Clench, CM).

Michigan. Upper Peninsula: Delta Co. 16.vi {per MCN). Gogebic Co.: 5 mi NE
Marenisco 2.vi.l967 {leg. J. H. Masters, MHM). Mackinac Co.: 12.vi (per MCN). Mar-
quette Co. {per MCN). Schoolcraft Co. 24.v-2.vi {per MCN). Lower Peninsula: Allegan
Co. {per MCN). Arenac Co. 4-21. v {per MCN). Barrien Co. 20.iv-v {per MCN). Arenac
Co. 4-21 V (per MCN). Cheboygan Co. 12-3 l.v {per MCN); Duncan Bay 12.V.1921
(Voss, 1954); Douglas Lake 29v.l939 (Voss, 1954). Crawford Co. 11-31.V (per MCN).
Emmet Co. (Moore 1960). Grand Traverse Co. 18.v (per MCN). Huron Co. 14.v (per
MCN); Sand Point 13.v. 1933 {leg. W. S. MacAlpine, CNC). Iosco Co. 26-30.V (per MCN).
Lake Co. 6-28. v. (per MCN). Montcalm Co. 7.v-10.vi (per MCN). Montmorency Co.
24-3 l.v (per MCN). Newaygo Co. 7.v-2.vi (per MCN). Oceana Co. 28. v (per MCN).
Ogemaw Co. 18.v (per MCN). Oscoda Co. 18-3 l.v (per MCN); Luzerne 3 l.v. 1 937,
30. V. 1939 (both leg. G. W. Rawson, CNC). Otsego Co. 17.v-4.vi (per MCN). Presque
Isle Co.: nr. Grand Lake 14.v. 1951 (Voss, 1954). Roscommon Co. 24-28. v (per MCN).
Wexford Co. 14.v (per MCN).

Minnesota. Beltrami Co. Bemidji 30.V.1967 (Huber, 1967). Chisago Co.: Taylors
Falls 20.V.1896 {leg. H. Skinner, ex ANSP, CM), 20.V.1967 {leg. J. H. Masters, JHM).
Goodhue Co.: Redwing (per JHM). Hennepin Co.: Minneapolis 16.v {leg. H. Skinner,

52

Annals of Carnegie Museum

VOL. 52

ex ANSP, CM). Houston Co.: 26.V.1940 (Macy and Shepard, 1941); Reno {per JHM).
MilleLacs Co.: Mille Lacs Lake 14.v. 1938 (Macy and Shepard 1941). Pine Co.: St. Croix
State Park ll,13.v.l965 (Huber, 1965), 16.V.1966) (Huber, 1966; also per JHM), 17-
24. V. 1967 (Huber, 1967). Washington Co.: Stillwater 12.v. 1934; Lakeland Junction
17.V.1938 (both Macy and Shepard, 1941). Watonwan Co.: Butterfield {leg. W. Linnsch-
eid, per JHM). Winona Co.: Dresbach 20. iv. 1938 {leg. ?, CM; also cf. Amhold, 1953);
Winona (per JHM), 20. v. 1937 (Macy and Shepard, 1941), 2.vi.l967.

Missouri. Benton Co.: Warsaw 26.iv.1962, 24.iv.1966 {leg. J. R. Heitzman, DAS,
PAO). Clay Co.: Liberty (William Jewell Campus) iv (1954-1958, leg. J. H. Masters,
JHM). Franklin Co.: Meramec Highlands 1 2.iv. 1 903 {leg. ?, CM), cf. also Amhold (1 953),
Meiners (1957). Jackson Co.: Blue Springs 22.iv.i965, 2. v. 1966 {leg. J. R. Heitzman,
PAO). Jefferson Co.: Cedar Hill 1 7.iv. 1 938 {leg. ?, OAS). St. Louis Co.: St. Louis 1 8.iv. 1 902,
12.iv.l903 (resp. CM, CAS), “seems to have become extinct at this locality” (Knetzler,
1912); Eureka 13,28.iv.l936 {leg. ?, CM ejc W. E. Sieker); Ranken 16.iv.l936 {leg. H.
1. O’Bryne, CNC), 26.iv (nfd, CM), cf. also Amhold (1953), Meiners (1957).

Montana. Carter Co.: nfd. 2.vi.l937 {leg. J. L. Kay, CM). Custer Co.: Miles City
16.V.1897 {leg. H. Skinner, CNC), also (Elrod, 1906). Gallatin Co.: Bozeman (Elrod,
1 906). Not located: Long Pine Hills 2.vi. 1937 {leg. J. L. Kay, CM); Aldridge (Elrod 1 906);
“M.C.” [=Miles City?] {leg. ? [perhaps H. Skinner], ex: ANSP, CM).

Nebraska. Cass Co.: Plattsmouth 13.v (Leussler, 1938). Dodge Co.: nfd (Leussler,
1938). Douglas Co. Omaha 4.v, 24. v (Leussler, 1938). Lancaster Co.: Lincoln 29. iv. 1939
{leg. W. H. Wagner, CNC). Otoe Co.: Nebraska City 10,20.iv.l958 {leg. J. A. Ebner,
CAS; and L. D. Miller, CM).

North Dakota. Morton Co.: 25. v (Puckering and Post, 1960). Slope co.: 24. v, l.vi
(Puckering and Post, 1960).

Oklahoma. Alfalfa Co.: Carmen 18.iv. 1954 {leg. D. L. Shorter). Major Co.: Orienta
9.iv.l944 {leg. Stallings and Turner, ex ANSP, CM). Payne Co.: Yale 30.iv.l958 {leg.
P. McCoy, OSU); Stillwater 12.iv.l941, leg. ?, OSU), 2.V.1960 {leg. C. J. McCoy, OSU).
Woodward Co.: Woodward 2. v. 1953; Boiling Springs St. Park 8.iv. 1954 (both leg., D.
L. Shorter, NWS). Note: All records, except that for Major Co., furnished by C. J. McCoy,
Jr.

Pennsylvania. Bedford Co.: 1 mi N Inglesmith [nr. Purcel] lO.v.1964 {leg. L. D.
Miller and H. K. Clench, CM). Not located: “Southwestern Pennsylvania” (Holland
1931; Macy and Shepard, 1941). Original source of information unknown to us. “Penn-
sylvania” (Klots, 1951; prob. from preceding authors).

South Dakota. Brookings Co.: Volga (Tmman, 1897).

Texas. “Western Texas” [=vic. Archer Co., teste F. M. Brown, letter to Clench
10.vi.l967] leg. Boll (Edwards, 1882; types of rosa, CM). Burnett Co.: Marble Falls (SS);
8.iv.l965 {leg. J. R. Heitzman, per ROK); nfd (larvae, leg. R. O. Kendall, per ROK).
Carson Co.: White Deer 25.iv.1943 {leg. H. A. Freeman, CNC); Skellytown {leg. H. A.
Freeman, per ROK). Dallas Co.: Dallas (SS); Garland {leg. H. A. Freeman, per ROK);
Irving {leg. H. A. Freeman, per ROK). Smith Co.: 4.iv.l964 {leg. C. A. Kendall, per
ROK.) Tarrant Co.: Ft. Worth (SS); 2nd week of March {per ROK). Travis Co.: Bull
Creek {leg. Sullivan, per ROK). Tyler Co.: Tyler St. Park, nr. Tyler 6.iv.l965 {leg. J. R.
Heitzman, PAO).

Virginia. Frederick Co.: W. Cross Jet., 24.iv.1938 {leg. A. H. dark, MNH); Whitacre,
9.V.39 (A. H. Clark, MNH); Lee Co.: White Shoals distr., 4 mi WSW Jonesville, 25-
28.iv. 79,9.v.82 (W. H. Wagner, Jr.).

West Virginia. Hampshire Co.: Ice Mountain, 19.iv.38, 7.V.39 {leg. A. H. Clark,
MNH); Forks of Cacapon, 8.V.39, 26.iv.41 (A. H. Clark, MNH); 7 mi NE Slanesville,
2,3.v and lO.v.1958 (leg. J. Bauer, CM); 3.1 mi N Purgitsville 15.iv. 1967 {leg. H. K.
Clench, CM). Hardy Co.: 1.8 mi S Baker 16.iv.l967 {leg. H. K. Clench, CM). Kanawha
Co.: Coalburgh 21,22.iv.l890; “Kana” [Kanawha] (leg. W. H. Edwards, including the

1983

Clench and Opler— Studies of Euchloe

53

types of olympia, CM). Mineral Co.: 4 mi W Burlington 29.iv.1967 {leg. H. K. Clench,
CM). Pendleton Co.: Reddish Knob 27.V.1981 {leg. P. A. Opler, MNH).

Wisconsin. Adams Co.: 1 3.v. 1 962 {ex W. E. Sicker, CM). Chippewa Co.: Lake Wissota,
nr. Chippewa Falls (Amhold, 1953); Chippewa Falls 5. v. 1957, lO.v. 1958 {ex W. E.
Sicker, CM). Dane Co.: Pine Bluff {ex W. E. Sicker, CM). Marinette Co.: Crivitz 1 6.v. 1958,
30“31.v.l957, 9.vi.l957 {leg. L. Griewisch, ex W. E. Sicker, CM). Marquette Co.: West-
field 19.V.1934 {leg. G. Heid, CAS). Vilas Co.: 7 mi W Eagle River 2.vi.l967 {leg. J. H.
Masters, JHM); 2 mi S Presque Isle 2.vi.l967 {leg. J. H. Masters, JHM). Waushara Co.:
Lake Lucerne [ca. 7 mi N Neshkoro, Marquette Co.] 27,29.v.l967 {leg. W. E. Sicker,
CM).

Wyoming. Converse Co.: vi.l900 (nfd, CM).

Literature Cited

Arnhold, F. R. 1953. Notes on collecting Anthocaris midea and Euchloe olympia.
Lepid. News, 6:99-100.

Beutenmuller, W. 1 898. Revision of the species of Euchloe inhabiting America north
of Mexico. Bull. American Mus. Nat. Hist., 10:235-248, ill.

Brooks, G. S. 1942. A revised checklist of the butterflies of Manitoba. Canadian Ent.,
74:31-36.

Brown, F. M., D. Eff, and F. Rotger. 1956 (1954-1956). Colorado butterflies. Proc.

Denver Mus. Nat Hist., 3-7:viii + 1-368 pp.

Clark, A. H., and L. F. Clark. 1951. The butterflies of Virginia. Smithsonian Misc.
Coll., 116(7):vii ± 1-239 pp.

Cook, C. 1948. Early spring collecting in Kentucky. Lepid. News, 2:22.

DeFoliart, G. R. 1956. An annotated list of southeastern Wyoming Rhopalocera.
Lepid. News, 10:91-101.

Edwards, W. H. 1882. Descriptions of new species of butterflies found in the United
States. Papilio, 2:45-49.

Elrod, M. J. 1906. The butterflies of Montana. Univ. Montana, Missoula, xvi +175

pp.

Field, W. D. 1938. A manual of the butterflies and skippers of Kansas. Bull. Univ.
Kansas, 39:238 pp.

Freeman, H. A. 1959. Butterfly collecting in Texas and New Mexico. J. Lepid. Soc.,
13:89-93.

Hanham, a. W. 1900. Additions to the list of Manitoba butteflies, with notes on other
species. Canadian Ent., 32:365-366.

Holland, W. J. 1931. The Butterfly Book (rev. ed.). Doubleday, Doran, Garden City,
New York, xii + 424 pp.

Howe, W. H. 1974. The butterflies of North America. Doubleday & Co., New York.
Huber, R. L. (compiler). 1966. Minnesota 1965 annual butterfly summary. Published
by author, St. Paul, Minnesota, 1 1 pp., hectographed.

. 1967. Minnesota collecting summary. Published by author, St. Paul, Minne-
sota, 29 pp., photolithed.

. 1968. Minnesota collecting summar>L Mid-Cont. Lepid., 2:1-24.

Klots, B. a. 1951. A field guide to the butterflies. Houghton Mifflin, Cambridge,
Massachusetts, xvi + 349 pp.

Knetzler, a. 1912. Observations on the Lepidoptera of St. Louis, Missouri, and
vicinity during 1911. Ent. News, 23:203-207.

Leussler, R. a. 1938. An annotated list of the butterflies of Nebraska (Lepid.: Rho-
palocera) [part 2]. Ent. News, 49:76-80.

Lindsay, S. A. 1971. A host plant for northern populations of Euchloe olympia (Pier-
idae). J. Lepid. Soc., 25:64.

54

Annals of Carnegie Museum

VOL. 52

Macy, R. W., and H. H. Shepard. 1941. Butterflies. Minneapolis, Univ. Minnesota
Press, ix + 247 pp.

Meiners, E. P. 1939. The life history of Euchloe olympia Edwards, with some notes
on its habits. Proc. Missouri Acad. Sci., 4:154-156.

. 1957. Lepidoptera collecting at Ranken, Missouri. Lepid. News, 10: 163-168.

Moore, S. 1960. A revised annotated list of the butterflies of Michigan. Occas. Papers
Mus. Zool., Univ. Michigan, 617:1-39.

Opler, P. a. 1966. Studies on Nearctic Euchloe. Parts 1 and 2. J. Res. Lepid., 5:39-
50.

. 1974. Studies on Nearctic Euchloe. Part 7. J. Res. Lepid., 13:1-20.

Platt, R. B. 1951. An ecological study of the mid-Appalachian shale barrens and of
the plants endemic to them. Ecol. Monogr., 21:269-300.

Puckering, D. L., and R. L. Post. 1960. Butterflies of North Dakota. Special Publ.,
Dept. Agr. Ent., North Dakota Agr. Coll., 32 pp.

Shull, A. 1907. Life history and habits of Anthocharis {Synchloe) olympia Edw. Ent.
News, 18:73-82.

Simmons, S. 1 957. Notes on ten new butterfly records for the state of Maryland. Lepid.
News, 10:157-159.

Truman, P. C. 1897. Lepidoptera in South Dakota [part 2]. Ent. News, 8:27-29.

Voss, E. G. 1954. The butterflies of Emmet and Cheboygan Counties, Michigan with
other notes on northern Michigan butterflies. Amer. Midland Nat., 51:87-104.

Wagner, W. H., Jr. 1977. A distinctive dune form of the marbled white butterfly,
Euchloe olympia (Lepidoptera: Pieridae) in the Great Lakes area. Great Lakes Ento-
mol., 10:107-112.

Wood, E., Jr., and C. W. Gottschalk. 1942. The butterflies of Roanoke and Mont-
gomery Counties, Virginia. Ent. News. 53:143-146, 146-164, 191-197.

» !

ISSN 0097-4463

ANNALS

(^CARNEGIE MUSEUM

CARNEGIE MUSEUM OF NATURAL HISTORY

4400 FORBES AVENUE • PITTSBURGH, PENNSYLVANIA 15213
VOLUME 52 30 MARCH 1983 ARTICLE 3

FURTHER RECORDS OF BATS FROM THE CENTRAL AFRI-
CAN REPUBLIC (MAMMALIA: CHIROPTERA)

J. E. Hill*

Abstract

These notes supplement the detailed and comprehensive survey of the known bat
fauna of the Central African Republic by Schlitter et al. (1982), with records of four
further species hitherto unreported from that country.

Introduction

Bats from the Central African Republic have been reviewed in detail
by Schlitter et al. (1982). These authors have added numerous species
to the hitherto poorly known bat fauna of the Republic and have
provided a list of previously recorded species not represented in the
collections that they reported, together with a prognosis of others that
although not yet recorded from the Republic may well be expected to
occur there. However, while their study was in course of publication,
a number of bats collected in the Republic by Dr. C. A. Spinage of the
Food and Agriculture Organisation of the United Nations was donated
to the British Museum (Natural History). For the most part they rep-
resent species already recorded by Schlitter et al. but the collection
includes four that are new to the reported fauna of the Central African
Republic and one other known from there apparently only by one
previous record. All come either from the Bamingui-Bangoran National
Park or from Bamingui and its environs, in the southern part of the

' Address: Department of Zoology, British Museum (Natural History), Cromwell Road,
London SW7 BSD, United Kingdon.

Submitted 17 September 1982. '

■\r'0

55

56

Annals of Carnegie Museum

VOL. 52

country. This brief account follows Schlitter et al. in treatment; mea-
surements have been made with dial calipers and are in millimetres.

Accounts of Species

Family Pteropodidae
Myonycteris torquata (Dobson, 1878)

1878. Cynonycteris torquata Dobson, Cat. Chiroptera Brit. Mus. pp. 71, 76, pi. 5, fig.
1. Angola.

Specimen ex^2m//2^^/(l).--Bamingui-Bangoran National Park, 400 m, 7°55'N, 19°26'E,
(1)(13).

Remarks. — This species was reported by Vielliard (1974:977) from
the “sud de la Republique Centre Africaine,” more specifically La
Maboke according to Bergmans (1976:195).

Measurements.— of forearm, 58.1; condylobasal length, 30.6;
alveolar length of maxillary toothrow (C-M^), 1 1.8.

Family Hipposideridae

Hipposidews cy clops (Temminck, 1853)

1853. Phyllorrhina cyclops Temminck, Esq. Zool. Cote de Guine, p. 75. River Boutry,
Ghana.

Specimen examined (l).~Bamingui River, Bamingui-Bangoran National Park, 400
m, 7°34'N, 19°40'E, (1) (?).

Remarks.— Xccovding to Schlitter et al. (1982:153), there is no pre-
vious record of H. cyclops from the Central African Republic but these
authors remarked that among others it is a species that might be ex-
pected to occur there. It is known from Senegal and Guinea-Bissau to
the southern Sudan and Kenya.

Measurements. — TQYigXh. of forearm, 68.4; condylobasal length, 24,8;
condylocanine length, 24.6; alveolar length of maxillary toothrow (C-
M^), 10.0.

Family Vespertilionidae

Nycticeius schlieffenii (Peters, 1860)

1860. Nycticejus schlieffenii Peters, Monatsber. K. Preussischen Akad. Wiss. Berlin, p.
223. Cairo, Egypt.

Specimen cxam/>2Ci/(l). — Bamingui-Bangoran National Park, 410 m, 7°33'N, 19'’58'E,

Remarks. -SchMner et al. (1982: 153) remarked of N. schlieffenii that
it was to be expected in the Central African Republic, although not
then recorded from that country. It is otherwise extensively distributed
from Mauretania and Ghana to Egypt, the Sudan, Somalia, and south-

1983

Hill— Bats from Central African Republic

57

western Arabia, thence through much of eastern Africa to Mozambique,
South Africa, Botswana, and Namibia.

Measurements. — 'Length, of forearm, 30.2; condylobasal length, 12.3;
alveolar length of maxillary toothrow (C-M^), 4.5.

Family Molossidae

Tadarida (Chaerephon) ansorgei (Thomas, 1913)

1913. Nyctinomus ansorgei Thomas, Ann. Mag. Nat. Hist., ser. 8, 11:318. Malange,
northern Angola.

Specimen examined (1).— Foubouloulou, 22 km W of Bamingui, (1) (19).

Remarks.— is no previous record of T. ansorgei from the Cen-
tral African Republic, although the species is known (Eger and Peterson,
1979:1890, fig. 2) to occur in Cameroon, the southern Sudan, and
Ethiopia, extending through eastern Zaire, Uganda, and Tanzania to
Mozambique, Zimbabwe, and Angola. This specimen agrees with T.
ansorgei as it is defined by Eger and Peterson (1979:1887) and with
the small sample of this species in the collections of the British Museum
(Natural History).

Hayman and Hill (1971:65) placed ansorgei in the nominate sub-
genus Tadarida {Tadarida), an allocation supported by Koopman (1975:
419, 424). More recently. Freeman (1981:109, 150) has transferred
ansorgei to the group of species included under Chaerephon, a change
with which Koopman (Freeman, 1981:1 50, footnote) concurs. Freeman
(1981) also considered Tadarida and Chaerephon generically distinct.

Measurements. — CQngXh. of forearm, 45.6; length of second phalange
of fourth digit, 10.4; condylobasal length, 18.2; interorbital width, 4.1;
height of braincase, 6.5; mastoid width, 11.1; alveolar length of max-
illary toothrow (C-M^), 7.2.

Otomops martiensseni (Matschie, 1897)

1897. Nyctinomus martiensseni Matschie, Arch. f. Naturg., 63, sect. 1, p. 84. Magrotto
Plantation, near Tanga, Tanzania.

Specimen — Bamingui-Bangoran National Park, 400 m, 7°55'N, 19°29'E,

(1)(19).

Remarks.— specimen from the Central African Republic rep-
resents a considerable westward extension to the northern part of the
distribution of this species, which has been known hitherto to extend
from Djibouti and Ethiopia through Kenya, northeastern and southern
Zaire to Tanzania, Natal, Malawi, Zimbabwe, and Angola; it also occurs
on Madagascar.

Measurements. — CQXigXh of forearm, 67.6; condylobasal length, 27.0;
alveolar length of maxillary toothrow (C-M^), 10.8.

58

Annals of Carnegie Museum

VOL. 52 I

Other species \

Besides the specimens noted above as being of particular interest, j
the collection made by Dr. Spinage also includes examples of Epo- \
mophorus gambianus, Micropteropus pusillus, Coleura afra, Tapho- I
zous mauritianus, Nycteris hispida hispida, Lavia frons affinis, Rhin- j
olopus fumigatus, R. landeri, Hipposideros ruber, H. abae, Pipistrellus I|
nanus, Eptesicus guineensis, E. somalicus, Scotophilus dinganii, S. ni- ;
gritellus, and Tadarida (Chaerephon) nigeriae. All of these taxa were I
reported from the Central African Republic by Schlitter et al. (1982). :

Literature Cited

Bergmans, W. 1976. A revision of the African genus Myonycteris Matschie, 1899
(Mammalia, Megachiroptera). Beaufortia, 24:189-216. I

Eger, J. L., and R. L. Peterson. 1979. Distribution and systematic relationship of
Tadarida bivittata and Tadarida ansorgei (Chiroptera: Molossidae). Canadian J. i
Zool., 57:1887-1895. ;

Freeman, P. W. 1981. A multivariate study of the family Molossidae (Mammalia, |
Chiroptera): morphology, ecology, evolution. Fieldiana, Zool., new ser., 7:vii +1- ^
173. I

Hayman, R. W., AND J. E. Hill. 1971. Order Chiroptera. Part 2: 1-73, m The mammals '
of Africa: an identification manual (J. Meester and H. W. Setzer, eds.), Smithsonian
Institution Press, Washington, D.C. i

Koopman, K. F. 1975. Bats of the Sudan. Bull. Amer. Mus. Nat. Hist., 154:355-443. ;
Schlitter, D. A., L. W. Robbins, and S. A. Buchanan. 1982. Bats of the Central ?

African Republic (Mammalia: Chiroptera). Ann. Carnegie Mus., 51:133-155.
ViELLiARD, J. 1974. Les Chiropteres du Tchad. Revue Suisse Zool., 81:975-991.

Z 73

ISSN 0097-4463

ANNALS

of CARNEGIE MUSEUM

CARNEGIE MUSEUM OF NATURAE HISTORY

4400 FORBES AVENUE * PITTSBURGH, PENNSYLVANIA 15213
VOLUME 52 30 MARCH 1983 ARTICLE 4

NEW BRACHIOPODS FROM THE GILMORE CITY
LIMESTONE (MISSISSIPPIAN) OF
NORTH-CENTRAL IOWA

John L. Carter

Curator, Section of Invertebrate Fossils

Abstract

Highly fossiliferous coquinoid lenses within the "^Cyathophyllum zone“ or uppermost
Gilmore City Limestone have produced extremely diverse collections of small inverte-
brates. The brachiopod faunule from these collections consists of at least eighteen species,
three of which are new and described herein. Two of the species are the basis for new
and unusual genera. lowarhynchus mirandum, new species, is the type species of a new
tiny smooth rhynchonellid, and Gerkispira spinosa, new species, is the type species of a
new impunctate costate spinose spiriferid. Setigerites facetus, new species, represents an
early occurrence of a typically Visean genus.

Introduction

The Gilmore City Limestone of north-central Iowa is a very light to
medium gray oolitic limestone with some interbedded hard tough gray
to blue dolomitic limestone and minor shale. The dolomitic beds and
shales occur only in the lower intervals but comprise much of the
exposed portion of the formation found in outcrop near the type section
at Gilmore City, Pocahontas County, Iowa.

Most of the major invertebrate groups of the Gilmore City Formation
of Iowa have been studied to some extent. The crinoids were first
described by Laudon (1933), corals by Carlson (1964), conodonts by
Anderson (1969), brachiopods by Carter (1972), and gastropods by
Harper (1977).

Submitted 22 September 1982.

^^|\\THSUA//,i|yy

ACC / v2; O

i ■- ’ lUOU

59

60

Annals of Carnegie Museum

VOL. 52

Table \.—List of brachiopods from the upper beds of the Gilmore City Limestone at
Hodge’s Quarry, Humboldt County, Iowa.

Rhipidomella sp. (juveniles?)

Hemiplethorhynchus subovatum Carter (3 mature specimens)
lowarhynchus mirandum, new genus and new species
Schuchertella sp. (juveniles?)

Rugosochonetes sp. (1 specimen)

Avonia sp. (1 specimen)

Setigerites facetus, new species
Pustula sp. (1 specimen)

? Merista sp. (internal morphology unconfirmed)
Cleiothyridina tenuilineata (Rowley)

Composita cf. C athabaskensis Warren (4 valves)

Composita sp. (juveniles?)

Prospira sp. (1 specimen)

Gerkispira spinosa, new genus and new species
Syringothyris sp. (2 incomplete specimens)

Punctospriifer sp. (juveniles?)

Cranaena sp. (2 adult specimens)

Girtyella sp. (3 specimens)

This formation crops out only in north-central Iowa but is reported
in subsurface across southwestern Iowa, southern Nebraska, northern
and western Kansas. Surface exposures in Iowa are limited to stream
banks and limestone quarries, the latter providing the best fossil ma-
terials. The quarries in the vicinity of the type section near Gilmore
City have produced many of the specimens upon which most of these
prior faunal studies were based. My earlier brachiopod paper (1972)
was based exclusively on such collections. Almost all of those speci-
mens were recovered from the lower beds of the formation exposed at
or near the type section.

The taxa that are the subject of this paper all were collected from
the uppermost “zone” or lithologic unit of London (1933), the “Cy-
athophyllum (=Vesiculophyllum) zone.” This unit is poorly exposed
and not fossiliferous in the quarries near Gilmore City, Pocahontas
County, but is very well exposed in a relatively new quarry just north
of Dakota City, Humbolt County. Originally named Hodge’s Quarry,
the property is now owned and operated by the P & M Stone Company.
This quarry originally consisted of two main pits, east and west. A thin
very fossiliferous coquina occurs within the oolitic beds about 35 ft
below the top of the formation in the west pit. A very large and diverse
collection of invertebrates was assembled from the coquina lenses by
A. J. Gerk. Smaller collections made by J. A. Harper of the Pennsyl-
vania Geological Survey and myself, plus the Gerk collections, pro-

1983

Carter— New Brachiopods from Iowa

61

vided the brachiopod specimens that form the basis for the present
study. The east pit was much less fossiliferous but a fine collection of
a new productid species was collected from a horizon 22 ft below a
distinctive marker bed of orange-brown shale a few feet below the top
of the formation. According to Gerk and Levorson (1982:69), the two
pits have become merged into a single large quarry in recent years and
virtually all of the highly fossiliferous material has been removed.
Brachiopods are very saprsely distributed at Hodge’s Quarry except
for these coquina lenses. The purpose of this paper is to describe several
interesting new brachiopods from Hodge’s Quarry, including two un-
usual new genera.

Most of the brachiopods from the coquina lenses of the west pit are
very small specimens. There is a mixture of small adults and juveniles
that reflect size sorting, probably from wave and current action in a
high energy shallow water environment.

Laudon (1933:29) collected six species of brachiopods from the “Cv-
athophyllum zone” at Humboldt, Humboldt County. None of these
species can be identified with certainty in the Hodge’s Quarry collec-
tions.

The Hodge’s Quarry Fauna

At least 18 species of articulate brachiopods occur in the Hodge’s
Quarry collections. Only two of these occur in the lower horizons or
at other localities. Hemiplethorhynchus subovatum Carter is common
in the lower beds of the formation at Gilmore City and is much less
common in Laudon’s ""Streptorhynchus zone.” Three specimens were
collected at Hodge’s quarry by J. A. Harper. Setigerites facetus, new
species, first occurs in the ""Streptorhynchus zone” at the quarries near
Gilmore City where it is very rare. At Hodge’s Quarry it is rare in the
lower beds but occurs in large numbers at a single horizon high in the
section. Most of the other 16 species (Table 1) are not reported from
the lower beds of the formation exposed in Pocahontas County. Many
are not identifiable to species level for lack of specimens with adult
characters.

Biostratigraphic Relationships

In my earlier Gilmore City Limestone paper (1972), I concluded that
the type Gilmore City brachiopod fauna could not be correlated pre-
cisely with either Late Kinderhookian or Early Osagian Faunas in the
Mississippi Valley region. This conclusion was essentially in agreement
with that of Laudon (1933) who instead readily correlated the Gilmore
City crinoids with Lodgepole Limestone crinoid faunas of the Cordil-
leran Region. Laudon arbitrarily assigned the Gilmore City Limestone

62

Annals of Carnegie Museum

VOL. 52

to the Late Kinderhookian because it lacked conclusively Osagian cri-
noid genera. Laudon’s crinoids nearly all came from the lower beds of
the formation exposed near the type section.

Recently, Harper (1977) also considered the Gilmore City Ls. to be
Latest Kinderhookian in its outcrop area. His conclusion was based
on his study of the Hodge’s Quarry gastropods but he admitted that
his judgement was arbitrary and that an Early Osagian correlation was
possible. Conodont evidence for precise correlation is scanty (Ander-
son, 1 973). Again, a Late Kinderhookian age is suggested in the absence
of Early Osagian taxa. No conodont evidence is available from the
"‘‘Cyathophyllum zone” of Laudon, the highest beds in the formation.

The brachiopods of the ''Cyathophyllum zone” at Hodge’s Quarry
for the most part provide little new biostratigraphic information. One
possible exception is the small athyridid Cleiothyridina tenuilineata
(Rowley). This distinctive little species was originally described from
the lower Burlington Limestone of Pike County, Missouri. The beds
from which the type specimens were recovered are of Early but not
Earliest Osagian age. However, this species also occurs in the Chappel
Limestone of central Texas and the “middle” Banff Formation of Al-
berta. The Chappel Limestone specimens occur in beds assignable to
the Gnathodus punctatus zone of Hass (1959), or roughly to the Si-
phonodella cooperi hassi— Gnathodus punctatus zone of Thompson and
Fellows (1970). The Banff specimens of C. tenuilineata all occur in the
Siphonodella cooperi hassi— Gnathodus punctatus zone of Baxter and
von Bitter (in press). Wherever recognized, this zone has been consid-
ered to be of latest Kinderhookian age. Thus, C. tenuilineata occurs
only in Late Kinderhookian and Early Osagian strata.

None of the other brachiopod species from Hodge’s Quarry offers
much assistance in accurately delimiting the age of the "" Cyathophyllum
zone.” The available evidence from all groups, such as it is, still suggests
a Late Kinderhookian or Early Osagian age for the Gilmore City Lime-
stone. Definitive data pointing to one series or the other are still lacking.

Systematic Paleontology

Order Strophomenida Opik
Superfamily Productacea Gray
Family Tolmatchoffiidae Sarycheva
Genus Setigerites Girty
Setigerites facetus, new species
Plate 1 (figs. 1-20)

1972. Buxtonial sp. B, Carter, p. 480, pi. 1, fig. 19.

Hoiotvpe.-cymn 34405.

Paratypes. -CMlStn 34402-34404, 34406, 34407.

1983

Carter —New Brachiopods from Iowa

63

Table 2.— Measurements of types (in mm) of Setigerites facetus, new species.

CMNH no.

Length

Width

Height

Surface

measure

Pedicle valves

34402

26.3

29.4

16.3

47.0

34403

28.4

28.3

18.6

49.1

34404

29.3

25.2

19.8

52.6

34405

26.8

25.2

17.1

49.0

Brachial valves (natural molds)

34406

21.0

26.7

__

—

34407

20.7

24.8

—

Description.— Ay QVdLgQ size for genus, outline transversely to longitudinally subovate,
lateral profile strongly gibbous; maximum width attained well anterior to hingeline;
visceral region subquadrate to convex, brachial valve geniculate, forming moderately
thick body cavity; trails moderately produced.

Pedicle valve almost evenly convex in lateral profile, being slightly more convex in
region of umbo and visceral disc; venter moderately convex, flattened, or rarely with
weak shallow sulcus; flanks dropping steeply to lateral margins; beak small, incurved,
slightly overhanging hingeiine; anterior margins of largest specimens flaring slightly;
radial ornament consisting of elongate spine ridges in umbonal region and irregularly
continuous fine costae over remainder of valve that increase by both intercalation and
bifurcation, with about 10-- 13 per cm near anterior margin; concentric ornament con-
sisting of weak irregularly spaced rugae posteriorly, forming weakly reticulate effect on
umbo and ears; spine bases irregularly scattered on crests of costae over most of valve;
few spines around sides of beak and near hingeline erect and very fine; spine bases on
ears and flanks near ears large, not numerous, forming sparse brush in most specimens
at cardinal extremities; growth lines very fine, sinuous, irregularly spaced; interior not
observed.

Brachial valve with moderately concave visceral disc; trail short, sharply geniculated;
ears clearly delineated by convex flexures; dorsum weakly concave, flanks more sharply
flexed near margins; pronounced small concave area produced near postero-medial mar-
gin by flexures that delimit ears; tiny elongate (protegulal?) node externally marking base
of cardinal process; radial ornament consisting of elongated dimples on much of visceral
disc, forming furrows between fine discontinuous costae anteriorly; dimples round and
large on ears; concentric ornament consisting of numerous weak irregularly spaced sin-
uous rugae and very fine growth lines; spine bases erect, fine, sparsely scattered on visceral
disc and trail.

Brachial valve interior with short stout sessile bilobed cardinal process, dorsally in-
clined from plane of visceral disc; median septum thick at base of cardinal process,
tapering anteriorly, extending forward at least to half length of visceral disc; lateral ridges
extending along hingeline almost to ears; other internal details not observed.

Measurements are given in Table 2.

Distinguishing characters.— This species is characterized by its lack
of, or poorly developed, flared rim or gutter on the pedicle valve trail,
a ventral umbo of moderate breadth, about 1 2 ribs per centimeter on
the ventral trail, the conspicuously dimpled dorsal visceral disc, the

64

Annals of Carnegie Museum

VOL. 52

1

1983

Carter— New Brachiopods from Iowa

65

sparse brush of relatively coarse spines on the ears of the pedicle valve,
and very weakly reticulate visceral disc.

Comparisons.— species that appear to be similar to this new
species are Dictyoclostus sciotoensis Hyde, 1953, from the Logan For-
mation of Ohio, and Setigerites lichwiniformis Sarycheva, 1963, from
Tournaisian strata of the Kuznets Basin. Both of these species share a
similar outline, profile, and ornament with S. facetus. Dictyoclostus
sciotoensis differs in being larger, its ears are more poorly defined,
especially on the visceral disc of the brachial valve, and it has slightly
finer costae with more numerous bifurcations on the trail of the pedicle
valve. Setigerites lichwiniformis can be distinguished by its poorly de-
fined ears on the brachial valve, slightly finer more continuous costae
with fewer spine ridges on the umbones, and it has a more dense brush
of spines at the cardinal extremities.

Smaller specimens of Productus jasperensis Warren, 1932, from the
Banff Formation of Alberta are similar to S. facetus. They can be
differentiated by their broader ventral umbo, more strongly reticulate
visceral disc, and dense spine brush on the ears. Mature specimens of
P. jasperensis are much larger and have a flared marginal rim or gutter.

Setigerites setigerus (Hall) and S. altonensis (Norwood and Pratten)
are substantially younger species. Both have more continuous costae
on the umbones and both have narrower more elongate ventral um-
bones that substantially overhang the hingeline. In addition, S. setig-
erus has very dense spine brushes on the ears and S. altonensis has
much finer costae than does S. facetus, new species.

Comments.— AssignmQni of this new species to Setigerites, a com-
mon Visean genus, is based on general similarity in external form and
ornament and close similarity in internal cardinalia. However, there
are possibly significant external differences between S. setigerus (Hail),
the type species, and S. facetus, new species. The type species virtually
lacks reticulation on the visceral disc in either valve, except near the
ears. It possesses a well developed gutter or concave flange around the
anterolateral margin, and it has more or less continuous costae that

^ PLATE 1

All figures XI

Figs. \-20.— Setigerites facetus, new species, 1-16, ventral, lateral, anterior, and posterior
views of four large pedicle valves, CM 34403-34406, including the holotype, figs. 9^12;
1 7-20, ventral and anterior views of two natural molds of brachial valve exteriors, CM
34406, 34407.

All specimens are from Hodge’s Quarry, east pit, 22 ft below a thin marker bed of
orange-brown shale near the top of the Gilmore City Limestone.

66

Annals of Carnegie Museum

VOL. 52

extend virtually to the hingeline in both valves, lacking the elongate
spine ridges seen on the visceral disc of S. facetus.

Setigerites facetus on the other hand has weak but distinct reticulation
on both valves, a gutter has not been observed in any specimen, al-
though the lateral slopes do flare slightly near the margin, and the
costae originate from elongate spine bases on the umbo of the pedicle
valve and from the complementary ornament of dimples on the bra-
chial valve.

Dictyoclostus sciotoensis Hyde and Productus jasperensis Warren,
discussed above, are species also assignable to Setigerites that possess
some or all of these different characters with S. facetus and it is possible
that these three species are closely related, perhaps representing an
early adaptive radiation of this genus into distinctly different environ-
ments.

Stratigraphic occurrence.— This species is very rare in the ''Strep-
torhynchus zone” of Laudon (1933) in the vicinity of the type section,
Pocahontas County, Iowa. At Hodge’s Quarry it is rare in the lower
beds of both pits, but is common at a single horizon in the east pit
about 22 ft below the orange-brown shale marker bed near the top of
the formation.

Order Rhynchonellida Kuhn
Superfamily Rhynchonellacea Gray
? Family Pontisiidae Cooper & Grant
lowarhynchus, new genus

Type species. — lowarhynchus mirandum, new genus and new species.

Species assigned. — Gtmxs monotypic.

Diagnosis. — small, smooth, elongate rhynchonellaceans with
subcardiiform to guttate outline; sulcus variably present in pedicle
valve, rarely in brachial valve; fold lacking; short dental plates present;
foramen hypothyridid, slitlike; deltidial plates lacking; dorsal septum
lacking; hinge plate often divided but inner hinge plates tending to
converge, rarely fused to form undivided hinge plate; crura thin, mod-
erately long, bladelike, falcifer, projecting antero-ventrally.

Comparisons.— Tiny entirely smooth rhynchonellaceans are rare in
the stratigraphic record. Dorsisinus Sanders is such a genus from the
Late Famennian Louisiana Limestone of Missouri. Sanders (1958:53)
mistakenly diagnosed Dorsisinus as possessing a divided hinge plate
and septalium by assuming that his silicified Mississippian species from
Mexico was conspecific with the type species, Centronella louisianensis
Weller. However, Dorsisinus louisianensis has an undivided hinge plate,
lacks a septalium, and has well developed crural plates that buttress
the hinge plate. Both Dorsisinus and the unnamed Mexican genus lack
a ventral sulcus; instead the venter is arched. The Mexican form does

1983

Carter —-New Brachiopods from Iowa

67

not seem to be related to the Iowa genus of comparable age, but Dor-
sisinus may indeed be a Devonian progenitor of lowarhynchus, new
genus.

Other smooth rhynchonellaceans that bear superficial resemblance
to lowarhynchus are Donella Rotai from the Lower Carboniferous of
the Donetz Basin and Acolosia Cooper and Grant from the Permian
of Texas. Donella lacks dental plates, possesses a dorsal septum, and
is of enigmatic affinities. Some species of Acolosia, especially A. elliptica
Cooper and Grant, are much more similar both externally and inter-
nally to the new Mississippian genus. Externally, Acolosia elliptica has
a similar outline and profile but differs in having a broader ventral
sulcus, a more acute ventral beak, and it never has a dorsal sulcus.
Internally, Acolosia has fairly wide outer hinge plates and the crura are
not as vertically compressed. Interestingly, both lowarhynchus and
Acolosia lack deltidial plates.

Discussion.— T\iQ lack of radial ornamentation and dorsal median
septum obviate ready assignment of the new genus to an established
Carboniferous rhynchonellacean family. The recently erected Permian
Family Pontisiidae of Cooper and Grant ( 1 976:20 1 9) includes the genus
Acolosia Cooper and Grant discussed above. Most external and internal
characters of lowarhynchus, new genus, more or less agree with Cooper
and Grant’s family diagnosis. However, most of the taxa assigned to
the family by Cooper and Grant have radial ornament of some sort
except for two species of Acolosia. There is also one significant internal
difference. Virtually all of the Permian species assigned to the Ponti-
siidae by Cooper and Grant have recognizable outer hinge plates,
whereas in lowarhynchus the crural bases arise directly under the socket
ridges, the outer hinge plates being completely suppressed. For these
reasons and because of the substantial hiatus in stratigraphic occur-
rences the new Mississippian genus is assigned to the Pontisiidae ten-
tatively.

lowarhynchus mirandum, new genus and new species
Plate 2 (figs. 1-44); Figs. 1 and 2

//o/o/v/?£^.-CMNH 34412.

Paratypes. -CMNH 34408-11, 34413-20.

Description. — \QTy small for the superfamily, subequally biconvex, with longitudinally
subovate, subcardiiform, or guttate outline; maximum width attained anterior to mid-
length; lateral profile lenticular to subovate; anterior commissure uniplicate to recti-
marginate; fold never developed; sulcus commonly present in pedicle valve, much less
commonly in brachial valve; ornament lacking except for irregularly spaced coarse growth
varices and very fine faint irregularly spaced growth lines; shell substance impunctate.

Pedicle valve moderately and evenly convex in profile in most specimens; some mature
specimens with sharply deflected growth lamellae at margins producing truncated profile;
lateral slopes moderately rounded or sometimes slightly flattened; maximum depth usually

68

Annals of Carnegie Museum

VOL. 52

Text-Fig. 1.— Transverse serial sections of lowarhynchus mirandum n. gen. n. sp., CM
34419, XI 2. The numbers refer to distance in mm from the ventral beak. This is a
mature specimen with a divided hinge plate, externally similar to the holotype, lacking
a dorsal sulcus.

1983

Carter— New Brachiopods from Iowa

69

Text-Fig. 2.— Transverse serial sections of lowarhynchus mirandum n. gen. n. sp., CM
34420, Xl2. The numbers refer to the distance in mm from the ventral beak. This is a
mature emarginate specimen with an undivided hinge plate and a sulcus in each valve.

attained near or slightly posterior to mid-length; umbonal region of variable breadth but
not inflated in lateral profile; beak small, nearly straight; beak ridges rounded; delthyrium
narrow, open, continuous with rounded hypothyridid foramen; sulcus, if present, orig-
inating anterior to umbonal region, shallow, rounded; venter usually flatened or weakly
convex in specimens lacking sulcus; dental plates short, thin, diverging slightly.

Brachial valve similar in depth and convexity to pedicle valve but with maximum
depth slightly anterior to that of opposite valve; umbonal region variably produced,
usually low and broad, never strongly inflated; beak inconspicuous; fold lacking, dorsum

70

Annals of Carnegie Museum

VOL. 52

Table 3.— Measurements (in mm) of types o/ lowarhynchus mirandum, new genus and

new species.

CMNH number

Length

Width

Thickness

34408

4.0

3.3

2.3

34409

3.9

3.5

2.5

34410

3.2

2.5

2.0

34411

2.5

2.0

1.5

34412

4.2

3.4

2.6

34413

4.1

3.4

2.5

34414

4.0

3.0

2.3

34415

3.1

2.4

1.6

34416

4.2

3.2

2.5

34417

3.9

3.0

2.0

34418

2.7

2.1

1.4

usually weakly convex, flattened, or with shallow sulcus, but only very rarely evenly
convex; sulcus, when present, very shallow, rounded, originating in dorsal umbonal region
as faint medial depression.

Dorsal interior lacking median septum; hingeplate usually divided, often with con-
verging inner hinge plates, rarely with fused single plate; socket ridges high, sturdy, stoutly
enclosing teeth; crural bases originating near dorso-medial edges of sockets with outer
hinge plates not recognizable; sockets deep and rounded; crura triangular or rodlike in
section near hinge plate, becoming comma-like anteriorly, then much flattened or com-
pressed vertically, bladelike, and projecting into ventral valve distally; weak low my-
ophragm present in most specimens.

Measurements are given in Table 3.

Comparisons. —This new species is not closely similar to previously
described rhynchonellaceans. Its small size variable sulcus and com-
plete lack of radial ornament are unusual in Mississippian strata, pos-
sibly unique. Dorsisinus louisianensis (Weller) is a tiny rhynchonella-
cean from the Louisiana Limestone (Late Famennian) of Missouri that
resembles I. mirandum, new species, in size and outline. Dorsisinus
louisianensis dilfers externally by invariably lacking a sulcus in the

PLATE 2 ->

All figures X6

Figs. \-AA. — lowarhynchus mirandum, new genus and new species, 1-16, ventral, dorsal,
anterior, and lateral views of four emarginate specimens with dorsal sulcus, CM 34408-
3441 1; 17-32, ventral, dorsal, anterior, and lateral views of four specimens with ventral
sulcus only, CM 344 1 2-344 1 5, including the holotype, figs. 1 7-20; 33-44, ventral, dorsal,
anterior, and lateral views of three rectimarginate specimens lacking a sulcus in either
valve, CM 34416-34418.

All specimens are from the thin coquina lenses in the upper 35 ft of the west pit at
Hodge’s Quarry, Humboldt County, Iowa.

1983

Carter— New Brachiopods from Iowa

44

72

Annals of Carnegie Museum

VOL. 52

pedicle valve, it usually has weak radial ribbing, and it has narrow
deltidial plates in some specimens. Inside the brachial valve of D.
louisianensis has substantially different cardinalia.

Stratigraphic occurrence. — Occurs only in the thin coquina lenses of ■
the upper oolitic beds of the Gilmore City Limestone at Hodge’s Quar-
ry, west pit. Many of the better type specimens were collected and
donated to the Carnegie Museum of Natural History by Arthur J. Gerk I
of Mason City, Iowa. |

Order Spiriferida Waagen |

Suborder Spiriferidina Waagen
? Superfamily Reticulariacea Waagen t

Family Uncertain !

Gerkispira, new genus '

Type species. — Gerkispira spinosa, new genus and new species. ji

Species assigned. — Type species only.

Diagnosis. SmuW transverse spiriferids with rounded lateral ex- j
tremities in all growth stages and poorly delimited fold-sulcus; both i
valves with numerous costae on flanks and fold-sulcus, those on flanks !'
being simple with rare exceptions, and those on fbld-sulcus occasionally
bifurcating; micro-ornament consisting of numerous fine hollow erect j
spines on costae and fine growth lines on entire surface; pedicle valve I
interior with slender dental adminicula; brachial valve interior with '

small striate cardinal process and short tabellae; shell substance im- |

punctate. j

Comparisons. — ImpuucXuXc entirely costate spiriferoids with erect !
hollow spines are rare in the paleontological record. Martynova (1961) ■

described such a genus, Spinospirifer, from the Famennian of Kazakh- |
Stan. Spinospirifer has a capillate lamellose micro-ornament, an un-
usual nasute fold-sulcus, acute lateral extremities, and thus, does not !
appear to be related to Gerkispira, new genus.

Comments.— The absence of endopunctae in Gerkispira is unex- I
pected. The spinose micro-ornament and spiriferoid costation would

PLATE 3 !

Figs. 1-19.-- Gerkispira spinosa, new genus and new species, 1 ,2, enlargements of micro- ;|
ornament showing broken spine bases of two typically preserved pedicle valves, CM
34392 (holotype) and CM 34393, X8; 3, enlargement of an unusually spinose brachial |

valve, CM 34399, X5; 4, 5, 15, 16, 18, 19, dorsal, and anterior views of three large *

branchial valves, CM 34396-34398, X2; 6, 7, 8, 9, ventral and anterior views of two
pedicle valves, CM 34393 and CM 34394, X2; 10-14, ventral, dorsal, lateral, anterior, ^
and posterior views of the holotype, CM 34392, X2; 17, enlargement of a small pedicle ;
valve, CM 34395, showing erect hollow spine bases, X8.

All specimens are from the west pit at Hodge’s Quarry, Humboldt County, Iowa.

1983

Carter— New Brachiopods from Iowa

73

74

Annals of Carnegie Museum

VOL. 52

ii

lead one to expect punctate shell structure. However, no punctae were
detected after careful examination of every specimen at hand. Fur-
thermore, other punctate taxa from the same collection, such as Cran~
aena sp., Punctospirifer sp., and Rhipidomella sp., have well preserved
punctae. It seems unlikely that the shell substances of Gerkispira, new
genus, has been altered.

This new genus is not closely similar to any previously described
genus and its familial affinities are uncertain. If one ignores the spinose
ornament it would be most similar to members of the Choristitidinae
by virtue of its rounded lateral extremities, numerous costae, poorly
differentiated fold-sulcus, and short tabellae. On the other hand spinose
ornamentation has not, to my knowledge, been reported in impunctate
Carboniferous Spiriferacea. The spinose non-capillate micro-orna-
ment, well rounded lateral extremities, poorly delimited fold-sulcus,
lack of true spiriferid denticulation (with taleolae in the secondary layer
of the interarea), and presence of short tabellae in the brachial valve
suggest reticulariacean affinities. However, strongly ribbed reticulari-
aceans are very rare. Two Triassic genera, Mentzeliopsis Trechmann,
and Spiriferinoides Tokuyama, appear to be reticulariaceans with
strongly costate flanks. Both possess hollow spines but they differ from
Gerkispira n. gen. in having a bald fold-sulcus and imbricate growth
lamellae externally, and both have a ventral median septum internally.
Although the strongly transverse outline, entirely costate macro-or-
nament, and erect spines are not reticulariacean characters, Gerkispira
might be best assigned to the Reticulariacea for the reasons cited above.

Gerkispira spinosa, new genus and new species
Plate 3 (figs. 1-19); Fig. 3

Holotvpe.-CMNYl 34392.

Paratypes. 39393-39401.

Description. ~^m2L\\tv than average for superfamily, subequally biconvex, transverse,
wider than long in all but earliest growth stages; outline spiriferoid; greatest width pos-
terior to mid-length; lateral extremities rounded in all growth stages; fold-sulcus mod-
erately developed, rounded, not well delimited from flanks; anterior commissure uni-
plicate; ornament consisting of numerous costae, irregularly spaced growth varices, fine
rounded spine bases usually on crests of costae, and fine regularly and closely spaced
growth lines; flanks with about 9 to 15 mostly simple costae, usually about 11 to 13,
sulcus with about 5 to 9 costae, with bifurcation usually restricted to sulcus-bounding
costae and median costa in posterior portion of valve; fold with about 7 to 1 2 costae.

Text-Fig. 3.— Transverse serial sections of Gerkispira spinosa n. gen. n. sp. The numbers
refer to the distance in mm from the ventral and dorsal beaks, respectively. A) a pedicle
valve, CM 34400. B) a brachial valve, CM 34401, showing the short tabellae.

1983

Carter —New Brachiopods from Iowa

75

76

Annals of Carnegie Museum

VOL. 52

Table 4.— Measurements (in mm) of Gerkispira spinosa, new genus and new species.

CMNH no.

Length

Hinge width

Maximum width

Thickness

Rib count

Holotype

34392

11.1

16.0

19.1

8.6

34

Pedicle valve

paratypes

34393

+8.4

8.5

±14.0

±4.0

28

34394

8.4

10.5

12.8

3.9

±32

34395

+5.0

-

+7.0

2.3

23

Brachial valve paratypes

34396

11.6

17.1

21.0

5.5

33

34397

9.3

14.0

16.9

4.0

32

34398

8.8

±15.0

17.7

3.8

34

34399

8.5

-

±15.2

3.8

26

usually 11 to 12, bifurcation usually restricted to median and one or two lateral sulcal
costae; total rib count ranging from about 27 to 39 per valve, usually about 32 to 34 in
larger specimens; intercostal furrows of moderate depth, rounded; erect hollow spine
bases arranged in radial rows on costae, very rarely occurring in intercostal furrows,
highly variable in density, appearing in earliest discernible growth stages; shell substance
thin, impunctate.

Pedicle valve moderately inflated, slightly thicker than brachial valve, forming low
equilateral triangle in posterior profile; maximum thickness attained in umbonal region,
just anterior to hingeline; umbonal region moderately developed; flanks slightly convex
medially, sloping evenly to antero-lateral margins; cardinal extremities slightly com-
pressed and rounded; beak small, wide, projecting posteriorly only slightly beyond dorsal
umbo; beak ridges subangular but only moderately well defined; interarea low, acutely
triangular, slightly procline, moderately concave, weakly vertically grooved in primary
shell layer; hingeline not denticulate; delthyrium about as high as wide, with short
outward-flaring stegidial plates; sulcus originating near beak, remaining narrow and
shallow in posterior half, becoming wider and rounded anteriorly but remaining shallow;
sulcus poorly delimited from flanks, with well rounded shoulders; sulcus-bounding costae
in umbonal region commonly bifurcating and becoming incorporated with sulcal slopes
anteriorly; moderate tongue produced; costae near cardinal extremities very faint; bi-
furcations of lateral costae extremely rare, especially anterior to umbonal region; origi-
nation of secondary sulcal costae sometimes by intercalation.

Pedicle valve interior with strong dental flanges and thin dental adminicula of moderate
length; subdelthyrial plate lacking; teeth small; muscle field weakly impressed; other
details not discernible in transverse section.

Brachial valve moderately to strongly convex, nearly as thick as opposite valve, with
maximum thickness attained at about mid-length of valve; umbonal region moderately
swollen, flanks evenly curving to lateral margins, but cardinal extremities slightly com-
pressed; beak small but well produced for this superfamily, projecting posteriorly to
slightly overhang delthyrial chamber; dorsal interarea low, slightly concave, acutely tri-
angular, orthocline to slightly anacline; fold originating anterior to umbonal region,
remaining low and rounded throughout, poorly differentiated from flanks except near
anterior margin; ornament similar to that of pedicle valve.

Brachial valve interior with small cardinal process composed of at least eight vertical

1983

Carter— New Brachiopods from Iowa

77

plates; inner socket ridges fused dorsomedially with crural bases, forming short tabellae;
muscle impressions obscure, no median ridge posteriorly; brachial details not observed.

Measurements are given in Table 4.

Stratigraphic occurrence. —This species is based on a single collection
of 93 specimens taken from the upper oolitic beds, about 35 feet below
the top of the formation, in the west pit at Hodge’s Quarry, near Dakota
City, Humboldt County, Iowa, SWV4, SWV4, Sec. 32, T. 92 N., R. 28
W. All of the better specimens were collected and donated to the Car-
negie Museum of Natural History by Arthur J. Gerk of Mason City,
Iowa. The genus is named for Mr. Gerk in appreciation of his out-
standing contribution.

Acknowledgments

The writing of this paper would not have been possible without the splendid collections
and generous contributions of Arthur J. Gerk of Mason City, Iowa. John A. Harper of
the Pennsylvania Geological Survey, Pittsburgh, contributed several fine specimens and
reviewed the manuscript, for which I am most grateful. Albert Kollar assisted in every
possible manner, but his serial sections of the tiny rhynchonellaceans merit special praise.
I thank Joaquin Rodriguez and Richard E. Grant for their useful suggestions for the
improvement of this paper.

Literature Cited

Anderson, W. I. 1969. Lower Mississippian conodonts from northern Iowa. J. Pa-
leontol., 43:916-^928.

. 1973. Mississippian conodonts in Iowa. Proc. Iowa Acad. Sci., 80:34-38.

Baxter, S., and P. H. von Bitter. In press. Conodont succession in the Mississippian
of southern Canada. Compte Rendu, IX International Congress of Carboniferous
Stratigraphy and Geology, Urbana, Illinois, 1979.

Carlson, K. J. 1 964. Corals of the Gilmore City Limestone (Mississippian) of Iowa.
J. PaleontoL, 38:662-666.

Carter, J. L. 1 972. Early Mississippian brachiopods from the Gilmore City Limestone
of Iowa. J. PaleontoL, 46:473-491.

Cooper, G. A., and R. E. Grant. 1976. Permian brachiopods of west Texas, IV.
Smithsonian Contr. Paleobiol. 24:2609-3159.

Gerk, A. J., and C. O. Levorson. 1982. Un usual beach deposits in oolitic carbonate
environments Mississippian and Recent. Proc. Iowa Acad. Sci., 89:68-70.

Harper, J. A. 1977. Gastropods of the Gilmore City Limestone (Lower Mississippian)
of north central Iowa. Unpublished doctoral dissert., Univ. Pittsburgh, Pittsburgh,
Pennsylvania, 317 pp.

Hass, W. H. 1959. Conodonts from the Chappel Limestone of Texas. U.S. Geol. Survey
Prof. Paper, 2944:365-399.

Hyde, J. E. 1953. Mississippian formations of central and southern Ohio. Ohio Geol.
Surv. Bull., 51:1-355.

Laudon, L. R. 1933. The stratigraphy and paleontology of the Gilmore City Formation
of Iowa. Univ. Iowa Studies Nat. Hist., 15(2): 1-74.

Martynova, M. V. 1961. Stratigrafiia i brakhiopody famenskogo iarusa zapadnoi
chasti tsentral’nogo Kazakhstana. Materiali po Geologii Tsentral’nogo Kazakhstana,
2:1-210.

Sanders, J. E. 1958. Brachiopoda and Pelecypoda. In Mississippian fauna in north-

78

Annals of Carnegie Museum

VOL. 52

western Sonora Mexico (W. H. Easton et al.), Smithsonian Misc. Coll., 1 19(3):41-
72.

Sarycheva, T. G., a. N. Sokolskaya, G. A. Besnossova, and S. V. Maksimova. 1963.
Brakhiopodi i paleogeografiia karbona Kuznetskoi kotlovini. Akad. Nauk SSSR,
Paleontol. Inst. Trudy, 95:1-547. j

Thompson, T. L., and L. D. Fellows. 1970. Stratigraphy and conodont biostratigraphy j
of Kinderhookian and Osagian (Lower Mississippian) rocks of southwestern Mis- !
souri and adjacent areas. Missouri Geol. Surv. Water Res. Report Invest., 45:1-
263. I

Warren, P. S. 1932. //? Allan, J. A., P. S. Warren, and R. L. Rutherford. A preliminary
study of the eastern ranges of the Rocky Mountains in Jasper Park, Alberta. Royal 1
Soc. Canada Trans. 26, series 3, section 4:225-248. i

ISSN 0097-4463

p7=- , /"S

ANNALS

of CARNEGIE MUSEUM

CARNEGIE MUSEUM OF NATURAL HISTORY

4400 FORBES AVENUE » PITTSBURGH, PENNSYLVANIA 15213
i VOLUME 52 17 JUNE 1983 ARTICLES

PHENETIC RELATIONSHIPS WITHIN THE CICONIIDAE

(AVES)

D. Scott Wood

Assistant Curator, Section of Birds

Abstract

The major skeletal elements of all 17 species of storks were analyzed using techniques
from multivariate statistics to assess the phenetic affinities within the avian family Cic-
oniidae. The common-part transformation and the use of taxonomic distances clustered
by the unweighted pair-group method with arithmetic averages gave significantly more
stable phenetic arrangements than other methods used. My results are very similar to
the classification of Kahl (1979Z>) at the generic level with the exception that Jabiru is
more similar to Ephippiorhynchus than would be indicated from Kahl’s classification.
Also, Kahl’s placement of these two genera in the tribe Leptoptilini is not supported by
results based on skeletal data; these genera are phenetically similar to the Ciconia species.

Introduction

The cosmopolitan avian family Ciconiidae is composed of the 17
species of storks (Table 1, nomenclature from Kahl 1971a, 1972a). By
1901, there was agreement as to the membership of the family except
for the Shoebill (Balaeniceps), which most authors considered to be in
a separate family (for example, Beddard, 1886, 1896, 1901; Garrod,
1875, 1877, 1878; Parker, 1860; Weldon, 1883) and is currently rec-
ognized as such (Kahl, 1979a). The classification of Peters (1931, Fig.
lA) summarized the systematic findings with the 17 species allocated
to 1 1 genera in two subfamilies. Prior to Peters’ classification, virtually
all studies involving storks either focused on the relationships of the
Ciconiidae as a whole to other birds (or vice versa) rather than on
relationships within the family or were concerned with describing a

Submitted 8 December 1982.

■j

79

80

Annals of Carnegie Museum

VOL. 52

A Peters (1931)

Subfamily Genus Species

Mycteria americana

— Ibis ibis

— — —Ibis leucocephalus

— — —Ibis cinerus

I Anastomus oscitans

•- Anastomus lamelligerus

— — Sphenorhynchus abdimii

Dissoura episcopus

[ Ciconia ciconia

* Ciconia nigra

Euxenura galeata

Xenorhynchus asiaticus

Ephippiorhynchus senegalensis

—Jabiru mycteria

[- Leptoptilos dubius

Leptoptilos crumeniferus

Leptoptilos javanicus

B Verheyen (1959)

Tribe Genus Species

Mycteria americana

Ibis cinereus
Ibis leucocephalus
Ibis ibis

Anastomus oscitans
Anastomus lamelligerus
Sphenorhynchus abdimii
Ciconia nigra
Ciconia ciconia
Euxenura galeata
Ephippiorhynchus senegalensis
Ephippiorhynchus asiaticus
Ephippiorhynchus mycteria
Dissoura episcopus
Leptoptilos javanicus
Leptoptilos dubius
Leptoptilos crumeniferus

Fig. 1. — Dendrograms representing (A) the classification of Peters (1931) and (B) the
relationships postulated by Verheyen (1959).

new feature of some particular species. More recent morphological
studies also have focused on the relationships of the family to other
birds (for example, Cottam, 1957, Ligon, 1967). With the exception
of Verheyen (1959), there has been no comprehensive investigation of

1983

Wood— Stork Phenetic Relationships

81

Table l. — Taxa and number of specimens used in this study (nomenclature from Kahl,

1972a).

Name

No.

specimens

Geographic range (Kahl, 19796)

Mycteria americana

5

Southern United States to Paraguay

American Wood Stork

Mycteria cinerea

2

Cambodia to Java

Milky Stork

Mycteria ibis

6

Africa, from Senegal to Sudan;

Yellowbilled Stork

south to Natal

Mycteria leucocephala

5

India and Sri Lanka to Vietnam

Painted Stork

Anastomus oscitans

5

India and Sri Lanka to Vietnam

Asian Openbiil Stork

Anastomus lamelligerus

4

Africa, from Senegal to Sudan;

African Openbiil Stork

south to Transvaal

Ciconia nigra

4

Much of Palearctic; South Africa

Black Stork

Ciconia abdimii

5

Africa, from Senegal to Uganda

Abdim’s Stork

and Yemen

Ciconia episcopus

5

India and Sri Lanka to Borneo

Woolynecked Stork

and the Philippines

Ciconia maguari

5

Colombia to Argentina

Maguari Stork

Ciconia ciconia

6

Europe to Turkistan; Cape Province;

White Stork

Eastern Siberia

Ephippiorhynchus asiaticus

5

India and Sri Lanka to northeastern

Blacknecked Stork

Australia

Ephippiorhynchus senegalensis

5

Africa, from Senegal to Sudan;

Saddlebill Stork

south to Transvaal

Jabiru mycteria

5

Mexico to Argentina

Jabiru Stork

Leptoptilos javanicus

5

Eastern India to Borneo

Lesser Adjutant Stork

Leptoptilos dubius

5

Northeastern India to Vietnam

Greater Adjutant Stork

Leptoptilos crumeniferus

6

Africa, from Senegal to Sudan;

Marabou Stork

south to Natal

the comparative morphology of members of the Ciconiidae, although
Furbringer (1888), Garrod (1873), and Vanden Berge (1970) included
representatives of the family in their comparative studies of muscles.
Verheyen (1959) examined a wide range of characters in an attempt

82

Annals of Carnegie Museum

VOL. 52 I

Kahl (1979)

Tribe Genus Species

I ■ » -t

Mycteria americana

Mycteria cinerea

— Mycteria ibis

pyfyQfQfjQ ieucocephaia

I Anastomus oscitans

' Anastomus iameiiigerus

Ciconia nigra

Ciconia abdimii

— Ciconia episcopus

Ciconia maguari

Ciconia ciconia

I Ephippiorhynchus asiaticus

* Ephippiorhynchus senegaiensis

— Jabiru mycteria

Leptoptiios javanicus

Leptoptiios dubius

— Leptoptiios crumeniferus

Fig. 2. — Dendrogram representing the classification of Kahl (1979b).

to clarify the relationships within the order Ciconiiformes (sensu Wet- |
more, 1960) and included all but one species of stork {Anastomus
oscitans) in his analysis. He proposed a classification for this family |
(Fig. 1 B) that differed in several important respects from that of Peters
(1931). In the process, he only reduced the number of genera to nine,
referring the genera Jabiru and Xenorhynchus of Peters to Ephippio- !
rhynchus and placing the genus Dissoura in the Leptoptilini, one of the !||
four tribes he recognized. I

Kahl (1966, \91\a, 1972a, \912b, 1972c \912d, 1972c 1973)
compared the ritualized courtship behavior of all 1 7 storks and con- '
siderably improved our knowledge of the intrafamilial relationships, i
His classification (Kahl, \919b) is shown in Fig. 2. Kahl also made j
important departures from both Peters’ (1931) and Verheyen’s (1959)
classifications, reducing the number of genera to six and allocating these ( |
genera to three tribes. Significantly, he combined several genera into i
Ciconia, merged Ibis into Mycteria and placed Anastomus and Myc-
teria in a separate tribe. He also associated Jabiru and Ephippiorhyn- ,
chus with Leptoptiios, although not on behavioral grounds (Kahl, 1973).

As part of a larger study involving the relationships of the Ciconiidae, !

I investigated the phenetic associations within the family as implied
by the structure of a number of different skeletal elements. I also use

1983

Wood— Stork Phenetic Relationships

83

Table I.— Bones used to infer phenetic relationships among the Storks; numbers of points
and characters defined on each bone.

Bone

Number of

Points

Linear

measurements

Angles

Ossa cranii, ossa faciei

16

39

51

Axis'

8

18

24

Humerus

17

36

57

Clavicula'

10

15

18

Coracoideum'

10

16

21

Sternum

9

21

33

Synsacrum, os coxae

11

21

30

Tibiotarsus'

12

20

24

T arsometatarsus

15

29

39

* Used only in preliminary analyses.

my empirical results to evaluate the efficacy of the phenetic methods
chosen, some of which have received little critical attention.

Materials and Methods

Skeletons of the 1 7 species were obtained (Table 1). Because of the paucity of material,
I measured the first five or six specimens encountered, as listed in Appendix I, regardless
of origin (unless a specimen was damaged); for some species fewer than five specimens
were available. I chose characters by: (1) comparing among a reference series of cico-
niiforms in the University of Oklahoma collections; (2) rejecting skeletal elements for
which I could discern only small variation over the series; (3) identifying points on each
of the remaining bones in such a way that for any individual I could identify the point
homologous to that of the reference series; (4) choosing interpoint distances that would,
when measured, describe variation visible to my eye. The interpoint distances were also
chosen so that they formed the sides of triangles; in this way I could calculate the angles
subtended by the sides of the triangles and thus have a relatively size-independent
measure of the shape of the bone. Using this procedure I chose 107 points spread over
nine bones or bone complexes. Fig. 3 shows the points chosen for each of the bones.
From this network I selected 215 measurements; these delimited 99 triangles giving 297
angular characters (Table 2). Descriptions of the points, characters and angles are listed
in Appendix II; nomenclature is taken from Baumel et al. (1979) and Bock and McEvey
(1969). The data for each bone were analyzed separately; in addition, I analyzed the total
data set (as defined below).

Computations were performed on the IBM 370-158 at the University of Oklahoma.
Some analysis was done using FORTRAN routines but most employed the packages
NT-SYS (Rohlf et al., 1979) and SAS (Barr et al., 1979). The three-dimensional models
were plotted using the program GRAFPAC developed by F. J. Rohlf

Each species was represented by the means of each character over all individuals.
Missing values were estimated using regression; of the approximately 12,000 measure-
ments used in this study, only 9 were coded as missing.

84 Annals of Carnegie Museum vol. 52

Fig. 3. — Representative skeletal elements of the storks showing the points defined in this
study. The characters defined from these points are listed in Appendix II. The specimens
illustrated are Leptoptilos sp. (UOMZ 7722, views 1-4), Ciconia ciconia (CM S-1813,
views 5-8, 10-15), Jabiru mycteria (UOMZ 6794, view 9), and Mycteria americana
(UOMZ 7720, views 16-23). The following bones and views are illustrated: (1) skull and
palate, ventral view; (2) skull and palate, left lateral view; (3) axis, right lateral view; (4)
axis, caudal view; (5) left humerus, proximal portion, cranial view; (6) left humerus,
proximal portion, caudal view; (7) left humerus, distal portion, cranial view; (8) left

Angles were calculated in radians in the following way: let A, B, and C be the sides j
of the triangle (the measured characters) and Abe the angle opposite A. Then X = 2*arctan ||
(R/{S - A)) where 5 = (vI + 5 + Q/2 and R = \/((S - A)*(S - B)*(S ~ Q)/VS. The i|
angles were calculated for all individuals and means were computed to represent each jj
species. Where, due to measurement error, angles could not be calculated (the sum of |
the lengths of two sides of a triangle was less than the third), the angles were estimated j j
by comparing the relative lengths of the sides and then assigning values representing ; j
angles of a very obtuse triangle of similar proportions. j {

Size (at least its linear effects) is considered by many workers to be a poor indicator i| |
of taxonomic relationships (Sneath and Sokal, 1973). For many taxonomic studies using j

1983 Wood— Stork Phenetic Relationships 85

humerus, distal portion, dorsal view; (9) clavicula, right dorsolateral view; (10) left
I coracoideum, dorsal view; (11) left coracoideum, ventral view; (12) sternum, left lateral
i view; ( 1 3) sternum, cranial view; ( 1 4) synsacrum, dorsal view; ( 1 5) synsacrum, left lateral
' view; (16) left tibiotarsus, proximal portion, dorsal view; (17) left tibiotarsus, distal
portion, dorsal view; (18) left tibiotarsus, proximal end; (19) left tibiotarsus, distal por-
tion, medial view; (20) left tarsometatarsus, proximal portion, dorsal view; (21) left
tarsometatarsus, proximal end; (22) left tarsometatarsus, distal portion, dorsal view; (23)
I left tarsometatarsus, distal portion, plantar view.

continuous characters, size has thus been considered a problem; using untransformed
‘ data, large (or small) species cluster together regardless of shape (see discussion in Sneath
I and Sokal, 1973:169-174). The problem of size has two facets: (1) the definition of size;
and (2) how to remove or reduce its effects. Sneath and Sokal (1973) suggested that if
in an ordination the first axis has mostly positive loadings then this axis represents a
! size axis. They proposed removing the effect by collapsing the ordination by that di-
j mension and using the remaining axes. Alternatively, they suggested removing a measure
I of size through the use of ratios, a common and widespread practice.

More recently Atchley et al. (1976), Atchley and Anderson (1978) and Atchley (1978)
have pointed out some potential dangers of using ratios. They suggested using regression

86

Annals of Carnegie Museum

VOL. 52

analysis to remove the effects of size (or some other variable). I expected size to have a
considerable influence on the magnitude of the raw measurements since the family
includes some of the largest of flying birds as well as a few of only moderate size (range
in height 76 cm-152 cm; Harrison, 1978). Principal component I (based on a matrix of
correlations among the characters from the total data set as defined below) did not
represent size since the high loadings were restricted to a specific set of characters and
were not all positive. However, clusters of species derived from the raw data (using the
unweighted pair-group method with arithmetic averages; UPGMA) appeared to reflect
a large size influence, that is, all large species clustered together. |

I operationally defined size as the mean of three length characters corresponding to ||j
the lengths of the tarsometatarsus, tibiotarsus and ischium (TA-DN, TI-EL, SY-CJ; see |j
Appendix II). To check that this variable reflected size, I clustered the species according
to size by eye and compared these groups to the UPGMA clusters formed using the size '

variable only. The results were concordant and were also very similar to clusters derived ■
from the raw data. I then regressed each variable against the size variable to obtain the v
regression estimate, subtracted the estimate from the original value and used the residual ;
as the character state. I

Explicitly, this procedure is as follows: let T be a vector of measurements for a given, |
character over all species. Let X be the size vector over all species. The estimate of T, I
(=T,) can be found from the equation Y = a + /3X, where a = Y - /3X and ^ c

Sx). The quantity r^y.x) is the correlation between the variables Y and X; Sy and % are f

the standard deviations of these variables. The residual, R, is found from = 7, — 7,. j
These residuals are uncorrelated with X (size). Size accounted for between 60% and 90% |:

of the variance of the data for the various bones. The angular data provided a second !|
transformation to reduce the effects of size; tests indicate the correlation with the size ||
variable to be much reduced. !|

For another transformation I removed the “common part” from each of the data sets. |j
This idea is described and developed more fully in Wood (1983a). Briefly, the variable 1
representing a particular stork (the vector of measurements on that stork) can be thought
of as having two parts: (1) a part predictable by regression from some reference variable
(some bird other than a stork) and (2) a part not predictable from the reference variable.

If the reference variable is chosen to estimate the essence of the stork family as closely
as possible, then I will refer to the first component as the common part. The second part
I will call the clustering part; the variance accounted for by the latter concerns differences
between the various storks as well as variation due to measurement error.

The clustering part is of most use when evaluating relationships within a group. The
common part can be removed in a manner analogous to the procedure outlined above
for removing size if an estimate of the common part can be found. In phenetic studies,
a species (or several species) that is very similar to the members of the group under study
can be used to estimate the common part; the more similar the outside species is, the j
better will be the estimate. For this study I used two species to estimate the common '
pavX—Balaeniceps rex and Scopus umbretta. They are osteologically very similar to the
storks and several authors have suggested they are related to the storks (for example, f
Mitchell, 1913; Parker, 1860; Peters, 1931; Wetmore, 1960). The regression equation j '
used in this case is: 7, = a + ftyX^ + ^2^2, where 7, is the estimate of all variables for «
species i, X^ and X2 are the vectors of character measurements for Scopus and Balaeniceps \
and a and ^ are as above (jd, and ^2 correspond to Scopus and Balaeniceps, respectively). '
Between 80% and 90% of the variance was removed from the various data sets by this
procedure. A separate estimate of the common part was also calculated from four other |
species (one pelican, one cormorant, and two New World vultures). Average taxonomic
distances (Sneath and Sokal, 1973) were calculated between all storks for each set of i ,
estimates for each bone. The average matrix correlation between estimates (over all J i
bones) was 0.91 indicating that the transformation (removal of the common part) is 1
robust to the choice of estimator for the common part. f

1983

Wood— Stork Phenetic Relationships

87

The use of means obscures variability within each species and, consequently, the
overlap in phenetic space. If substantial overlap exists, there may be insufficient infor-
mation present in a particular matrix to discern the relationships among the elements
(taxa). Partitions from such a data set should give less consistent taxonomic results than
partitions taken from a case showing much less overlap because in the first case the
variation is more nearly random with respect to the taxa. For all of the transformations
(size removed, angles, common part), I subjected each of the data sets to a multiple
discriminant function analysis (using PROC DISCRIM of SAS). My interest was in the
resolution afforded by the data: how many individuals were misclassified? If the resolution
was poor (that is, poor discrimination among species resulting in many misclassifications,
particularly if they were misclassified to species thought on other grounds to be relatively
dissimilar or unrelated) then I judged that using means would be likely to compound
errors and give discordant results. Of the nine bones originally used (Table 2), five showed
sufficient resolution to be useful in subsequent analyses (the remaining four produce far
less consistent results). Thus, the remainder of the analyses were restricted to the fol-
lowing: skull, humerus, sternum, synsacrum, and tarsometatarsus. One reason for the
poor resolution resulting from use of the other bones is that the original measurements
were devised to describe variation over all ciconiiforms (sensu Wetmore, 1960); the
elements not used further (axis, clavicula, coracoideum, and tibiotarsus) are very uniform
within the Ciconiidae but show considerable variation over the several families of the
Ciconiiformes.

All subsequent analyses (except as noted) were performed on each of the 1 8 data sets—
six partitions (skull = SK, humerus = HU, sternum = ST, synsacrum = SY, tarsometa-
tarsus = TA, all characters from these five bones = ALL) with three transformations for
each (size removed by regression = SIZE, size effect reduced by using angles = ANGLES,
common part removed = COMMON). A given data matrix is named by concatenating
the transformation code with the partition code (for example SIZE-SK, ANGLES-SY).
The following analyses were performed after standardization of the data: ( 1 ) a phenogram
summarized the results of a UPGMA cluster analysis on a matrix of average taxonomic
distances. (2) Factor analysis using a matrix of correlations among taxa (Q-type) was
performed with a secondary factor structure matrix obtained by oblique rotation using
the functionplane technique of Katz and Rohlf (1974). Communalities were estimated
using the maximum of the absolute values of the elements for each variable in the
correlation matrix and all factors with eigenvalues greater than 1.0 were retained. The
secondary structure matrix was interpreted as a taxon by character matrix and used to
produce inter-taxon correlations as the similarity measure. These were subjected to a
UPGMA cluster analysis and the results summarized in a phenogram. This procedure
seeks to identify patterns of common variation (clusters of taxa) and emphasize these
by rotating the axes to be coincident with the clusters. An oblique rotation was used
because there is no a priori reason to expect clusters of biological organisms to be
orthogonal. The analysis reconstitutes the among-taxa correlation matrix but emphasizes
the similarities between members of a cluster because the components of variance unique
to a given taxon are discarded. Thus, the phenograms closely resemble phenograms
produced by UPGMA clustering of the original inter-taxon correlation matrix except
that the clusters are better defined and phenograms based on different partitions of the
data are slightly more consistent (i.e., have a higher average matrix correlation among
results).

(3) Projections of the taxa on axes from a principal components analysis (the number
of components was determined by the number of eigenvalues greater than 1 .0) were
subjected to a cluster analysis using the adaptive hierarchical clustering scheme (AHCS)
of Rohlf ( 1 970); these results were summarized in a phenogram. AHCS can find elongated
clusters in multidimensional space rather than only the hyperspheroidal clusters searched
for by most clustering algorithms (including UPGMA). The degree of elongation sought
is controlled by a parameter of the program (set to 1.0 for this study). The technique

88

Annals of Carnegie Museum

VOL. 52

uses inter-taxon distances for the similarity measure and so may give results similar to
UPGMA clustering of these distances, particularly if elongated clusters do not exist.
Because of computer limitation, AHCS was not run on the combined (ALL) data sets.
(4) Non-metric multidimensional scaling (MDS) was used to reduce the dimensionality
of the data to three. The species were plotted on these axes using GRAFPAC to produce
three-dimensional (3-D) models; these give better visual representation of the distances
among taxa than is usually possible using phenograms. (5) Minimum-spanning trees for
taxa (Prim networks) were derived from average distance matrices and superimposed
on the 3-D models. (6) Matrix correlations were calculated between all pairs of pheno-
grams and between a phenogram and its basic similarity matrix (cophenetic correlations).

Codes for the three phenogram-producing analyses are: DIST (UPGMA clustering of
distances), AHCS (AHCS clustering of distances) and CORK (Q-type factor analysis).
Phenograms, basic similarity matrices and other results were named by concatenating
these codes with the codes for transformation and partition (e.g., COMMON-SK-DIST,
SIZE-ALL-CORR). A dendrogram of phenograms was created by applying UPGMA to
the matrix of correlations among phenograms (after changing the sign on the matrix
correlations between distance analyses and correlation analyses). Further discussion of
these techniques may be found in Sneath and Sokal (1973) or in the documentation to
NT-SYS (Rohlfet al., 1979).

Results and Discussion

Fifty-one phenograms were constructed to represent the similarities
among the storks. Fig. 4 is a dendrogram of these phenograms based
on correlations between all pairs of cophenetic matrices. The cophe-
netic correlation of this dendrogram is 0.693, an indication of some
distortion in the original associations. Figs. 5-7 show phenograms rep-
resentative of the major clusters present in the dendrogram (using an
arbitrary level of 0.4); additional phenograms are shown in Figs. 8-10
and in Wood (1982, 1983Z?).

Fig. 5A depicts the phenogram for ANGLES-HU-DIST, a represen-
tative of one of the two largest clusters in Fig. 4. Except for Anastomus
oscitans, which is well separated from the other species, the clusters
correspond closely to the genera of Kahl (1979 A Fig. 2). In addition
to A. lamelligerus being associated with Ciconia, Jabiru clusters with
the Ciconia species. This is one of only four phenograms to support
Kahl’s placement of Ephippiorhynchus with Leptoptilos (the others are
ANGLES-HU-AHCS, COMMON-HU-CORR and COMMON-SK-
CORR). The cophenetic correlation is 0.737 due to distortion among
the major branches of the phenogram.

The phenogram of COMMON-SK-AHCS (Fig. 5B) is a represen-
tative of the other large cluster in Fig. 4. The structure of this phenogram
matches the structure of Kahfs (1979Z?) classification (Fig. 2) closely
with the exceptions that Anastomus and Ephippiorhynchus are asso-
ciated with Ciconia, and Jabiru clusters with Mycteria.

In the phenogram for COMMON-ALL-CORR (Fig. 6 A), the species
cluster very tightly, a phenomenon generally true of the CORR analyses
because of the emphasis placed on close similarities by the factor an-

1983

Wood— Stork Phenetic Relationships

89

0.0

0.2

0.4

I

0.6

_J

0.8

I

1.0

icc = 0.693

Cc

COMMON-SK-DIST

ANGLES-SK-DIST

ANGLES-SK-AHCS

ANGLES-HU-AHCS

ANGLES-HU-DIST

COMMON-SK-CORR

SIZE-SK-DIST

SIZE-TA-AHCS

ANGLES-SK-CORR

SIZE-TA-CORR

SIZE-SK-CORR

SIZE-HU-CORR

ANGLES-TA-DIST

ANGLES-TA-AHCS

ANGLES-TA-CORR

SIZE-SK-CORR

SIZE-SY-DIST

SIZE-SY-AHCS

COMMON-SK-AHCS

COMMON-HU-AHCS

COMMON-ALL-DIST

COMMON-ST-DIST

COMMON-HU-DIST

COMMON-ST-CORR

COMMON-SY-DiST

SIZE-TA-DIST

ANGLES-SY-DiST

ANGLES-ST-DIST

ANGLES-ST-AHCS

SIZE-HU-DIST

SIZE-ALL-DIST

SIZE-HU-AHCS

COMMON-SY-AHCS

SIZE-SK-AHCS

COMMON-ALL-CORR

COMMON-HU-CORR

ANGLES-HU-CORR

ANGLES-ST-CORR

COMMON-TA-DIST

COMMON-TA-CORR

COMMON-TA-AHCS

SIZE-ST-CORR

SIZE-SY-CORR

COMMON-SY-CORR

ANGLES-SY-CORR

ANGLES-SY-AHCS

ANGLES-ALL-DIST

ANGLES-ALL-CORR

COMMON-ST-AHCS

SIZE-ST-DIST

SIZE-ST-AHCS

Fig. 4. — Dendrogram of phonograms derived from a UPGMA analysis of all possible
matrix comparisons between the cophenetic matrices for the 5 1 phonograms of this study
(scale is in correlation units).

90

Annals of Carnegie Museum

VOL. 52

ANGLES-HU-DIST

A

Mycteria americana
Mycteria cinerea
Mycteria ibis
Mycteria leucocephala
Anastomus lamelligerus
Ciconia abdimii
Ciconia episcopus
Ciconia nigra
Ciconia ciconia
Ciconia maguari
Jabiru mycteria
Ephippiorhynchus asiaticus
Ephippiorhynchus senegalensis
Leptoptilos javanicus
Leptoptilos crumeniferus
Leptoptilos dubius
Anastomus oscitans

0.4

rcc = 0.737

COMMON-SK-AHCS

'Mycteria americana
-Mycteria cinerea
-Mycteria ibis
-Mycteria leucocephala
-Jabiru mycteria
-Anastomus oscitans
■Anastomus lamelligerus
-Ciconia nigra
-Ciconia maguari
-Ciconia abdimii
-Ciconia episcopus

■ Ciconia ciconia
-Ephippiorhynchus asiaticus

- Ephippiorhynchus senegalensis
-Leptoptilos javanicus

■ Leptoptilos dubius
■Leptoptilos crumeniferus

Fig. 5. — Phenograms representative of the major clusters of the dendrogram of pheno-
grams: (A) derived from UPGMA clustering of taxonomic distances based on angular
characters of the humerus; (B) derived from an AHCS analysis of taxonomic distances
based on the COMMON transformation of skull characters. Scales of both phenograms
are in units of average taxonomic distance.

alytic procedure used. The composition of the clusters is similar to that
found in the previous examples with the exception of Ciconia which
is split among two groups. Jabiru is most similar to Ephippiorhynchus
and Ciconia, whereas Leptoptilos is closer to Mycteria. The cophenetic
correlation of 0.893 indicates relatively little distortion in the pheno-
gram.

The phenograms in Figs. 6B, 7A and 7B are representatives of rel-
atively small clusters of the dendrogram of phenograms (Fig. 4). The

1983

Wood— Stork Phenetic Relationships

91

-0.6 -0.2 0.2 0.6 1.0
I i I I j COMMON-ALL-CORR

Mycteria americana
Mycteria cine re a
Mycteria ibis
Mycteria leucocephala
Leptoptilos javanicus
Leptoptilos crumeniferus
Leptoptilos dubius
Anastomus oscitans
Anastomus lamelligerus
Ciconia abdimii
Ciconia episcopus
Ciconia nigra
Ciconia ciconia
Ciconia maguari
Ephippiorhynchus senegalensis
Ephippiorhynchus asiaticus
Jabiru mycteria

-0.6 -0.2 0.2 0.6 1.0

I I —J I 1 COMMON-SY-CORR

B

rcc = 0.715

C

T

d

Mycteria americana
Mycteria cinerea
Mycteria ibis
Ciconia maguari
Ciconia ciconia
Jabiru mycteria
Mycteria leucocephala
Ephippiorhynchus asiaticus
Ephippiorhynchus senegalensis
Leptoptilos javanicus
Leptoptilos dubius
Leptoptilos crumeniferus
Anastomus oscitans
Anastomus lamelligerus
Ciconia abdimii
Ciconia nigra
Ciconia episcopus

Fig. 6. — Phenograins representative of the major clusters of the dendrogram of phono-
grams. Both are derived from UPGMA clustering of correlations taken from a Q-type
factor analysis of characters transformed using the COMMON transformation: (A) based
on all characters; (B) based on synsacral characters. Scales of both are in correlation
units.

phenogram for COMMON-SY-CORR (Fig. 6B) shows several clusters
that are similar to Kahl’s genera but both Mycteria and Ciconia are
split. Mycteria leucocephala clusters with Ephippiorhynchus and C.
abdimii is associated with Anastomus. The cophenetic correlation (0.715)
is relatively low; the relationships of the storks implied by the phe-
nogram are somewhat distorted from the basic similarities. Mycteria
and Anastomus cluster together in the phenogram SIZE-ALL-DIST
(Fig. 7A). Four of the five species of Ciconia also cluster together but

92

Annals of Carnegie Museum

VOL. 52

2.0 1.6 1.2

0.8

SIZE-ALL-DIST

A

’’cc = 0.818

Mycteria americana
Mycteria cinerea
Mycteria ibis
Anastomus oscitans
Anastomus lamelligerus
Mycteria leucocephala
Ciconia abdimii
Ciconia episcopus
Ciconia maguari
Ciconia ciconia
Leptoptilos javanicus
Ciconia nigra
Jabiru mycteria
Ephippiorhynchus asiaticus
Ephippiorhynchus senegalensis
Leptoptilos dubius
Leptoptilos crumeniferus

2.0

ANGLES-TA-DIST

B

Mycteria americana
Mycteria cinerea
Mycteria leucocephala
Mycteria ibis
Anastomus oscitans
Anastomus lamelligerus
Leptoptilos dubius
Ciconia episcopus
Leptoptilos javanicus
Leptoptilos crumeniferus
Ciconia abdimii
Ciconia maguari
Ephippiorhynchus asiaticus
Jabiru mycteria
Ciconia nigra
Ciconia ciconia

Ephippiorhynchus senegalensis

= 0.730

Fig. 7. — Phenograms representative of the major clusters of the dendrogram of pheno-
grams: (A) derived from UPGMA clustering of taxonomic distances based on the SIZE
transformation of all characters; (B) derived from UPGMA clustering of taxonomic
distances based on angular characters of the tarsometatarsus. Scales of both are in average
taxonomic distance units.

Leptoptilos javanicus is separated from its congeners. Some distortion
is present (cophenetic correlation of 0.8 1 8). The phenogram ANGLES-
TA-DIST (Fig. 7B) represents a cluster in Fig. 4 composed of the three
analyses using the data set ANGLES-TA. These three phenograms are
the only results concordant with Verheyen’s (1959) suggestion that
Ciconia episcopus is closely related to Leptoptilos. Only the Mycteria

1983

Wood— Stork Phenetic Relationships

93

cluster resembles the groups proposed by Kahl {\919b, Fig. 2). The
cophenetic correlation of 0.730 is relatively low for this study.

Clearly, there are considerable differences between some phenograms
(the range of matrix correlations is —0.10 to 0.95). These differences
are not only between transformations and analyses but also between
data sets for a given transformation and analysis.

Rohlf and Sokal (1981) pointed out that methods that produce more
stable classifications (other factors being equal) would be preferred over
those that produce less stable classifications. They also noted that sta-
bility, by itself, is not a good criterion since perfect stability can be
achieved by making the results independent of the data. However, this
criterion can be useful when comparing methods chosen for reasons
other than stability (for example, to emphasize a particular aspect of
the data).

Rohlf and Sokal (1980, 1981) have discussed several definitions of
stability, of which robustness of a classificatory procedure to changes
in character sets (congruence of resultant classifications) is applicable
here. They defined congruence as “agreement of separate classifications
arrived at by the same algorithms and based on the same set of op-
erational taxonomic units but on different sets of characters” (Rohlf
and Sokal, 1981). The sets of characters should not represent different
classes of characters because such an evaluation of stability would be
confounded with an evaluation of the non-specificity hypothesis (Rohlf
and Sokal, 1980). Matrix correlations are a useful measure of the rel-
ative congruence of classifications although this measure may not be
optimal for evaluating absolute levels of concordance (Rohlf, 1982).

The partitions of the characters used in this study provide an op-
portunity to evaluate the stability of the methods employed. Both the
transformations of the data as well as the analyses applied to these data
are evaluated with respect to the stability of the resulting classifications.
Table 3 lists the average matrix correlations among phenograms over
all partitions (five bones) for ail transformations and analyses consid-
ered. Ranges and standard deviations are included. Within each trans-
formation and analysis (for example, ANGLES-DIST) the 10 corre-
lations used to calculate the mean are not independent since they
represent all possible comparisons among the five partitions. However,
at least four comparisons are independent and the degrees of freedom
in all significance tests are adjusted accordingly. Significant heteroge-
neity exists among the means listed in Table 3 (analysis of variance,
3.35, Fo 5[4 36] < 2.69). If the means are ranked, no significant dif-
ferences exist between adjacent means. However, the analysis COM-
MON-DIST has a significantly higher mean than any analysis except
COMMON-AHCS {t = 2.71, /0.5m = 1-86 [one tailed] between COM-

94

Annals of Carnegie Museum

VOL. 52

Table 3. —Statistics on matrix correlations between all pairs of cophenetic matrices de-
rived from each of the five data partitions. Each treatment of the data is analyzed sepa-
rately; for each analysis the number of possible intercorrelations is 10.

Analysis

Mean

Standard

deviation

Range

COMMON-DIST

.352

.182

.805-.291

COMMON-AHCS

.407

.193

.833-.177

COMMON-CORR

.353

.100

.471-.190

SIZE-DIST

.312

.145

.520-.051

SIZE-AHCS

.275

.151

.568-.107

SIZE-CORR

.339

.171

.679-.102

ANGLES-DIST

.303

.192

.597-. 102

ANGLES-AHCS

.187

.209

.667— .007

ANGLES-CORR

.291

.108

.475-.151

MON-DIST and COMMON-CORR). If the analyses are pooled so that
the dilferences between transformations ean be evaluated, then the
COMMON transformation gives significantly more congruent results
(that is, a higher mean) than does SIZE {t = 1.21 A, ^o.5[i6] 1.746 [one

tailed]). The ANGLE transformation gives results with lower average
congruence than SIZE. These findings are supported by a more detailed
analysis of a subset of the data which is presented elsewhere (Wood,
1983a).

The mean matrix correlation among partitions is very low for both
the SIZE and ANGLE transformations (Table 3). These transforma-
tions were chosen to emphasize the shape of the skeletal elements used
as the source of information about relationships among the storks.
Shape was defined in two different ways— in the SIZE transformation
shape was the residual variance remaining from a statistical removal
of a size estimate; in the ANGLE transformation shape was defined as
the angles subtended by pairs of measurements on the bones. The
emphasis on shape (or the de-emphasis of size) has been commonly
used in phenetic studies because the single characteristic “size” often
has such a pervasive influence as to mask the information about re-
lationships present in the other characters used in the study. However,
the classifications I have derived from the size transformations show
rather low stability (as measured by matrix correlations) regardless of
the analysis employed.

In contrast to findings of some other investigators (for example,
Schnell, 1970; Robins and Schnell, 1971; Hellack, 1976) but similar
to my earlier results on cranes (Wood, 1979), distances gave the most
stable (consistent) results. However, this may be more an effect of the
transformation than of other considerations since the transformations

1983

Wood— Stork Phenetic Relationships

95

2.0 1.6 1.2

COMMON-ALL-DIST

A

*”00 = 0.832

Mycteria americana
Mycteria cinerea
Mycteria ibis
Mycteria leucocephala
Anastomus oscitans
Ansatomus iameliigerus
Ciconia nigra
Ciconia ciconia
Ciconia abdimii
Ciconia episcopus
Ciconia maguari
Ephippiorhynchus asiaticus
Ephippiorhynchus senegalensis
Jabiru mycteria
Leptoptilos javanicus
Leptoptiios dubius
Leptoptilos crumeniferus

2.0

COMMON-SK-DIST

B

I'cc = 0.850

Mycteria americana
Mycteria cinerea
Mycteria ibis
Mycteria leucocephala
Ciconia nigra
Ciconia abdimii
Ciconia episcopus
Ciconia ciconia
Ciconia maguari
Jabiru mycteria
Ephippiorhynchus asiaticus
Ephippiorhynchus senegalensis
Leptoptilos javanicus
Leptoptilos dubius
Leptoptilos crumeniferus
Anastomus oscitans
Anastomus Iameliigerus

Fig. 8. — Phenograms derived from UPGMA clustering of average taxonomic distances
based on the COMMON transformation of the data: (A) all characters; (B) skull char-
acters. Scales of both are in units of average taxonomic distance.

used by the investigators cited above were quite different from those
used here.

Both analyses (UPGMA and ARCS) of distances among the taxa
using the COMMON transformation produced results that were more
stable than the CORK analysis but only the UPGMA (DIST) was
significantly higher. The DIST (UPGMA) and ARCS analyses are very
similar except that ARCS can also find clusters that are not hyper-
spheroidal. If the clusters are elongated then ARCS should better detect
them and, hence, give more stable results. This is not the case here

96

Annals of Carnegie Museum

VOL. 52

COMMON-HU-DIST

A

rcc = 0.859

Mycteria americana
Mycteria cinerea

Mycteria ibis \.

Mycteria ieucocephaia i

Anastomus oscitans I

Anastomus lameliigerus i,

Ciconia abdimii I

Ciconia episcopus

Ciconia nigra ,

Ciconia maguari |i

Ciconia ciconia i

Ephippiorhynchus asiaticus '

Jabiru mycteria j

Ephippiorhynchus senegalensis
Leptoptilos javanicus |

Leptoptilos crumeniferus
Leptoptilos dubius

2.0 1.6 1.2

COMMON-ST-DIST

B

fee “ 0.852

q

Mycteria americana !;

Mycteria cinerea |1

Mycteria ibis ['

Mycteria Ieucocephaia

C Anastomus oscitans

Anastomus lameliigerus i;

Ciconia nigra '

Ciconia abdimii I

Ciconia episcopus

Ephippiorhynchus asiaticus |

■Ciconia ciconia I.

■Ephippiorhynchus senegalensis li

Ciconia maguari

Jabiru mycteria f

Leptoptilos javanicus j

Leptoptilos dubius i:

Leptoptilos crumeniferus ||

Fig. 9. — Phenograms derived from UPGMA clustering of average taxonomic distances
based on the COMMON transformation of the data: (A) characters of the humerus; (B)
characters of the sternum. Scales of both are in units of average taxonomic distance.

although the dilference in stability (measured by matrix correlation) is ,
not significant. Inspection of the 3-D models (Figs. 1 1-16; these rep- I
resent the interspecific distances very closely) reveals little elongation i
of clusters.

Thus, for the purpose of revealing phenetic structure consistent over |
different sets of characters, the transformation and analysis COM-
MON-DIST is most effective. Figs. 8-10 show the phenograms from f
the six partitions (five bones plus the combined data) for this trans- *
formation and analysis. All except COMMON-SY-DIST are relatively f
good representations of the basic similarity matrices (cophenetic cor- I

1983

Wood— Stork Phenetic Relationships

97

2.0

COMMON-SY-DiST

A

1

1

frr =0.737

Mycteria americana
Mycteria cinerea
Mycteria ibis
Mycteria leucocephala
Anastomus oscitans
Anastomus lamelligerus
Ciconia nigra
Ciconia episcopus
Ciconia abdimii
Ciconia maguari
Ciconia ciconia
Jabiru mycteria
Ephippiorhynchus asiaticus
Ephippiorhynchus senegalensis
Leptoptiios javanicus
Leptoptilos crumeniferus
Leptoptiios dubius

2.0

COMMON-TA-DIST

B

Mycteria americana
Mycteria cinerea
Mycteria ibis
Mycteria leucocephala
Anastomus oscitans
Anastomus lamelligerus
Ciconia episcopus
Ciconia abdimii
Ciconia nigra
Ciconia ciconia
Ciconia maguari
Leptoptiios javanicus
Leptoptiios crumeniferus
Leptoptiios dubius
Ephippiorhynchus asiaticus
Ephippiorhynchus senegalensis
Jabiru mycteria

Fig. 10.— Phenograms derived from UPGMA clustering of average taxonomic distances
based on the COMMON transformation of the data: (A) synsacrum characters; (B)
tarsometatarsus characters. Scales of both are in units of average taxonomic distance.

relations greater than 0.830). The cophenetic correlation for COM-
MON-SY-DIST is 0.737, indicating distortion in the phenogram. Figs.
11-16 show the 3-D models derived from MDS analyses of the COM-
MON transformed data sets. These give more accurate representations
of the dissimilarities (distances) among the storks than the phenograms
but are much more difficult to interpret as hierarchically nested sets of
clusters (classifications in the usual sense). The minimum spanning
networks superimposed on the models give further information on the
close affinities of the storks.

98

Annals of Carnegie Museum

VOL. 52

COMMON-ALL-DIST

1. Mycteria americana

2. Mycteria cinerea

3. Mycteria ibis

4. Mycteria leucocephala

5. Anastomus oscitans

6. Anastomus lamelligerus

7. Ciconia nigra

8. Ciconia abdimii

9. Ciconia episcopus

10. Ciconia maguari

1 1 . Ciconia ciconia

12. Ephippiorhynchus asiaticus

13. Ephippiorhynchus senegalensis

14. Jabiru mycteria

15. Leptoptilos javanicus

16. Leptoptilos dubius

17. Leptoptilos crumeniferus

Fig. 1 1.— Three-dimensional model showing arrangement of storks in phenetic space
based on a multidimensional scaling analysis of all characters transformed using the
COMMON procedure. A minimum spanning network derived from taxonomic distances
calculated from the transformed data is superimposed on the model.

The six phenograms of Figs. 8-10 have many clusters in common
(the average matrix correlation between them is 0.531). In particular,
some of the genera proposed by Kahl (1972^?, 1979b) appear consis-
tently as clusters. The four species placed by Kahl in the genus Mycteria !
cluster together in all analyses, as do the two species of Anastomus and j
the three of Leptoptilos. Within the Mycteria association, M. ibis and '
M. leucocephala are a mutually close pair in all analyses except COM-
MON-SY-DIST (Fig. lOA). In the latter, M. ibis is the species most
similar to M. leucocephala but not vice versa (see Fig. 15). Mycteria f
americana and M. cinerea form a mutually close pair in three of the
analyses (COMMON-ALL-DIST, Fig. 8 A; COMMON-HU-DIST, Fig.

9 A; and COMMON-TA-DIST, Fig. lOB) but in the remaining three, i

1983

Wood— Stork Phenetic Relationships

99

1. Mycteria americana

2. Mycteria cinerea

3. Mycteria ibis

4. Mycteria ieucocephala

5. Anastomus oscitans

6. Anastomus iameUigerus
1. Ciconia nigra

8. Ciconia abdimii

9, Ciconia episcopus

10. Ciconia maguari

11. Ciconia ciconia

12. Ephippiorhynchus asiaticus

13. Ephippiorhynchus senegaiensis

14. Jabiru mycteria

15. Leptoptilos javanicus

16. Leptoptilos dubius

17. Leptoptilos crumeniferus

COMMON-SK-DIST

Fig. 12.— Three-dimensional model depicting an arrangement of storks in phenetic space
based on a multidimensional scaling analysis of skull characters transformed using the
COMMON procedure. A minimum spanning network derived from taxonomic distances
calculated from the transformed data is superimposed on the model.

M. americana is represented as the most divergent of the group. This
latter arrangement is not new; M. americana has traditionally been
separated generically from the other wood storks (Fig. lA).

The two species of Ephippiorhynchus cluster less consistently, being
a mutually close pair in only three of the phenograms (Figs. 8A, B and
10 A). However, E. senegaiensis is more similar to E. asiaticus than to
any other species in analyses of both the humerus and sternum (Figs.
13-14). For the tarsometatarsus data, E. senegaiensis is nearly as sim-
ilar to E. asiaticus as is Jabiru (Fig. 16).

The relationships of the species placed by Kahl into the genus Ciconia
are less consistently portrayed. In only two phenograms (Fig. 8A and
8B) are all five species clustered together to the exclusion of the other
storks. However, some close similarities are most consistently found:
C. abdimii and C. episcopus group together in all analyses and in four
cases are mutually close pairs; C. nigra, C. maguari, and C. ciconia
are present in the same cluster in four analyses. Table 4 summarizes
the nearest neighbors (along the minimum spanning network in the

100

Annals of Carnegie Museum

VOL. 52

a

c3

^ kJ

c ^

tko g

-S ^

c. I

-S ^

G 2

5 II

is

^ rN

O Co

^co

^ §

1^
^ II
•S b

<1

G to

^ II

Vi

V

X

V

<

H

o

&

H

GO

<

H <

,2

H

GO j-

5

.2

D 1

;dh

Kd

C/D

GO

a

a

S

hJ

GO

H <

1

a

s

8

8

&

>^

GO

D H

G

H

GO

c/3

H

GO

tmipqv

K, HU,

ALL

GO

GO

kJ

kJ

<

1

<

s;

H

AS

P

G

3

SS

a

•2

<D

G

<D

V

0^

Ciconia

nigra

■1 8

-a
~S

Q G

a

s

5

o

•2

G

I

1983

Wood— Stork Phenetic Relationships

101

COMMON-HU-DIST

1. Mycten’a americana

2. Mycteria cinerea

3. Mycteria ibis

4. Mycteria leucocephala

5. Anastomus oscitans

6. Anastomus lamelligerus

7. Ciconia nigra

8. Ciconia abdimii

9. Ciconia episcopus

10. Ciconia maguari

1 1 . Ciconia ciconia

12. Ephippiorhynchus asiaticus

13. Ephippiorhynchus senegalensis

14. Jabiru mycteria

15. Leptoptiios javanicus

16. Leptoptiios dubius

17. Leptoptiios crumeniferus

Fig. 13.— Three-dimensional model showing arrangement of storks in phenetic space
based on a multidimensional scaling analysis of humerus characters transformed using
the COMMON procedure. A minimum spanning network derived from taxonomic dis-
tances calculated from the transformed data is superimposed on the model.

multidimensional character space) of each of the Ciconia species for
each data set. For two species, the results are the same regardless of
data set— the most similar species to C. abdimii is always C. episcopus
and the most similar to C. maguari is in every case C. ciconia. The
converse is not true; that is, C. maguari is not always the most similar
species to C. ciconia. In fact, in only one of the six analyses (synsacrum)
is this true. However, in all but two of 30 cases the nearest (most
similar) species to a member of the genus Ciconia is a congener (C.
episcopus is most similar to Anastomus oscitans in the sternal analysis
and C. ciconia is most similar to Ephippiorhynchus asiaticus in the

102

Annals of Carnegie Museum

VOL. 52

COMMON-ST-DIST

Mycteria americana
Mycteria cinerea
Mycteria ibis
Mycteria leucocephala
Anastomus oscitans
Anastomus lameiligerus
Ciconia nigra
Ciconia abdimii

9. Ciconia episcopus

10. Ciconia maguari

1 1 . Ciconia ciconia

12. Ephippiorhynchus asiaticus

13. Ephippiorhynchus senegaiensis

14. Jabiru mycteria

15. Leptoptilos javanicus

16. Leptoptilos dubius

17. Leptoptilos crumeniferus

Fig. 14.— Three-dimensional model showing arrangement of storks in phenetic space
based on a multidimensional scaling analysis of sternum characters transformed using
the COMMON procedure. A minimum spanning network derived from taxonomic dis-
tances calculated from the transformed data is superimposed on the model.

analysis of tarsometatarsal data; see Figs. 14 and 16). C nigra is in-
termediate between the two pairs of species discussed above; C. ciconia
is most similar to it in four analyses (and vice versa in three of those)
and C episcopus is most similar to it in the remaining two (sternal and
synsacral data). These results suggest that for summarizing the phenetic
relationships among the Ciconia species using a phenogram, COM-
MON-ALL-DIST (Fig. 8A) is the best consensus. C. maguari should
be shown more similar to C. ciconia and C. nigra than is depicted.
The 3-D model for this analysis (Fig. 1 1) gives a good graphic summary
of the phenetic relationships of the Ciconia species.

The relationships of Jabiru are less consistent than for other storks.
In the phenograms it is associated with Ephippiorhynchus in COM-
MON-ALL-DIST (Fig. 8 A) and COMMON-TA-DIST (Fig. lOB), and
with Ciconia ciconia and/or C. maguari in COMMON-HU-DIST (Fig.
9A), COMMON-ST-DIST (Fig. 9B) and COMMON-SY-DIST (Fig.

1983

Wood— Stork Phenetic Relationships

103

COMMON-SY-DiST

1. Mycterla americana

2. Mycteria cinerea

3. Mycteria ibis

4. Mycteria ieucocephala

5. Anastomus oscitans

6. Anastomus lamelligerus
Ciconia nigra
Ciconia abdimii
Ciconia episcopus

10. Ciconia maguari
Ciconia ciconia

12. Ephippiorhynchus asiaticus

13. Ephippiorhynchus senegalensis

14. Jabiru mycteria

15. Leptoptilos javanicus

16. Leptoptilos dubius

17. Leptoptilos crumeniferus

Fig. 15. “Three-dimensional model showing arrangement of storks in phenetic space
based on a multidimensional scaling analysis of synsacrum characters transformed using
the COMMON procedure. A minimum spanning network derived from taxonomic dis-
tances calculated from the transformed data is superimposed on the model.

10 A); it is very different from other storks when only the skull is
considered (see Figs. 8B, 12). However, in the analyses using data from
either the humerus of sternum, the species most similar to Jabiru is
Ephippiorhynchus asiaticus (Figs. 13, 14). Jabiru thus shows closest
similarities to Ephippiorhynchus but is also clearly related to some
members of the Ciconia group.

The relationships among the groups discussed above (corresponding
to the genera of Kahl) also show some consistent features. Leptoptilos
is divergent from other groups in five of the six analyses, COMMON-
TA-DIST (Fig. lOB) being the exception, and in all but COMMON-
SK-DIST (Fig. 8B) is the most divergent group. In this latter exception
(Fig. 8B), Anastomus is the most divergent but because the two storks
of this genus have such highly modified bills and associated skull struc-
tures adapted to a particular feeding method (Kahl, 1971Z?), it is not

104

Annals of Carnegie Museum

VOL. 52

COMMON-TA-DIST

Mycteria americana
Mycteria cinerea
Mycteria ibis
Mycteria leucocephaia
Anastomus oscitans
Anastomus lamelligerus
Ciconia nigra
Ciconia abdimii
Ciconia episcopus
Ciconia maguari
Ciconia ciconia
Ephippiorhynchus asiaticus
Ephippiorhynchus senegalensis
Jabiru mycteria
Leptoptilos javanicus
Leptoptilos dubius
Leptoptiios crumeniferus

Fig. 16.— Three-dimensional model showing arrangement of storks in phenetic space
based on a multidimensional scaling analysis of tarsometatarsus characters transformed
using the COMMON procedure. A minimum spanning network derived from taxonomic
distances calculated from the transformed data is superimposed on the model.

surprising that an analysis of skull characters shows them to be quite
divergent. In four of the other five analyses, Anastomus is closely linked
with Mycteria\ in the fifth (COMMON-ST-DIST, Fig. 9B), the asso-
ciation is much less close but, as can be seen from the corresponding
3-D model (Fig. 14), Mycteria is more similar to Anastomus than to
any other genus or group. Members of the Ephippiorhynchus- Jabiru
group are associated with the Ciconia species in three of the phenograms
(Figs. 8A, 9) but in every 3-D model this group (or a majority of the
group) is linked in the minimum spanning network to Ciconia species.

Summary of Phenetic Results i

A major use of summaries of relationships is to predict the character |
states of characters outside those used in the study. Phenetic methods S
are designed explicitly to maximize this predictiveness. They are not I
designed, as are cladistic methods, to produce estimates of phylogeny. j

1983

Wood— Stork Phenetic Relationships

105

However, if in the evolution of the characteristics of the group under
study there is little homopiasy (convergence, parallelism) and rates of
divergence are approximately equal in all lineages, the phenetic and
cladistic summaries will be very similar.

I used phenograms and 3-D models as summaries of the phenetic
relationships among the storks. A further form of summary is a hier-
archic classification of names such as used in the Linnaean system.
Phenograms can easily be transformed into such classifications since
both are composed of hierarchically nested sets. The 3-D models do
not correspond directly to such classifications but often exhibit greater
accuracy in portraying the phenetic relationships. Each of the consid-
erable number of phenograms and 3-D models constructed in this study
emphasizes a different aspect of the phenetic relationships. Clearly, if
the characters whose states are to be predicted all relate to a specific
portion of the animal, then a summary' emphasizing that part is likely
to be most useful. For example, predictions of the similarities in feeding
habits and apparatus of the storks would likely be most accurately
obtained from a phonogram, 3-D model, or classification emphasizing
characteristics of the head region of the skeleton. However, for a general
statement, a summar>' representing the similarity over all characters is
most appropriate provided it shows little distortion of the consensus
of relationships implied by the different partitions of the data.

Both the technique employed to examine the relationships and the
character set used affect the results produced. For the data analyzed in
this paper, the COMMON transformation is clearly preferred as is the
use of average distance as a similarity measure. UPGMA is appropriate
because there is little evidence for elongated clusters among the storks.
The phonogram resulting from this analysis on all characters is COM-
MON-ALL-DIST (Fig. 8A). This appears to represent the consensus
of relationships well with the exception that Ciconia maguari should
be shown more similar to C. ciconia and C nigra. A better summary
but one that is more difficult to compare with previous classifications
(the most common form of summary of relationships) is given by the
3-D model for this analysis (Fig. 1 1).

My results are very similar to the relationships implied by KahFs
{\919b) classification (Fig. 2) except that Jahiru is more similar to
Ephippiorhynchus than would be predicted from KahFs summary. Also,
KahFs placement of these two genera in the tribe Leptoptilini is not
supported by results based on skeletal data; these genera are pheneti-
cally similar to the Ciconia species. Specific taxonomic recommen-
dations and further discussion of the relationships among the storks is
in Wood (1983/)).

106

Annals of Carnegie Museum

VOL. 52

Acknowledgments

This research was supported by a grant from the Frank M. Chapman Memorial Fund
of the American Museum of Natural History. I am grateful to the following curators
who allowed me to use specimens in their care: John W. Fitzpatrick, Field Museum
Natural History; Ned K. Johnson, Museum of Vertebrate Zoology, University of Cali-
fornia, Berkeley (through the aid of grant BMS 7200102 from the National Science
Foundation); Robert M. and Marion J. Mengel, Museum of Natural History, University
of Kansas; Raymond A. Paynter, Museum of Comparative Zoology, Harvard University;
Lester L. Short, American Museum of Natural History; Charles G. Sibley, Peabody
Museum, Yale University; Robert W. Storer, Museum of Zoology, University of Mich-
igan; Richard L. Zusi, National Museum of Natural History. The following individuals
helped considerably in the recording and checking of data: Phyllis J. Aldag, Kathy Allison,
Barbara Frantz, Christine E. Wood, Christopher S. Wood, Darwin L. Wood, and Law-
rence R. Wood. Karen M. S. Herbst, Nancy J. Perkins, and Christoph Malczewski
prepared the drawings for Fig. 3; Nancy J. Perkins prepared the remainder of the figures.
I owe a great debt to James R. Estes, Douglas W. Mock, Alan Nicewander, and especially
Gary D. Schnell for their assistance, review and patience during the tenure of this project.
Finally, I wish to thank my wife, Charlotte, for her support and assistance with most
aspects of the work. This research was conducted as part of the Ph.D. program at the
University of Oklahoma, Norman.

Literature Cited

Atchley, W. R. 1978. Ratios, regression intercepts, and the scaling of data. Syst. ZooL,
27:78-83.

Atchley, W. R., and D. Anderson. 1978. Ratios and the statistical analysis of bio-
logical data. Syst. Zook, 27:71-78.

Atchley, W. R., C. T. Gaskins, and D. Anderson. 1976. Statistical properties of
ratios. 1. Empirical results. Syst. Zook, 27:137-148.

Barr, A. J., J. H. Goodnight, J. P. Sale, W. H. Blair, and D. M. Chilko. 1979.
SAS— Statistical analysis system. SAS Institute, Raleigh, North Carolina.

Baumel, j. j., a. S. King, A. M. Lucas, J. E. Breazile, and H. E. Evans. 1979. Nomina
anatomica avium. Academic Press, New York, 637 pp.

Beddard, F. E. 1886. Notes on the convoluted trachea of a curassow {Nothocrax
urumutum), and on the syrinx in certain storks. Proc. Zook Soc. London, 1886:321-
325.

. 1896. Note upon Dissoura episcopus with remarks upon the classification of

the Herodiones. Proc. Zook Soc. London, 1896:231-235.

. 1901. Some notes upon the anatomy and systematic position of the ciconiine

gQmxs Anastomus. Proc. Zook Soc. London, 1901:365-371.

Bock, W. J., and A. McEvey. 1969. Osteology of Pedionomus torquatus (Aves: Ped-
ionomidae) and its allies. Proc. Roy. Soc. Victoria, 82:187-232.

CoTTAM, P. A. 1957. The pelecaniform characters of the skeleton of the Shoebill Stork,
Balaeniceps rex. Bulk British Mus. (Nat. Hist.), 5:49-72.

Furbringer, M. 1888. Untersuchungen zur Morphologie und Systematik der Vogel,
I. Specieller Theik T. J. Van Holkma, Amsterdam.

Garrod, a. H. 1873. On certain muscles of the thigh of birds and on their value in
classification. Part 1. Proc. Zook Soc. London, 1873:626-644.

. 1875. On the form of the trachea in certain species of storks and spoonbills.

Proc. Zook Soc. London, 1875:297-301.

. 1877. Note on an anatomical peculiarity in certain storks. Proc. Zook Soc.

London, 1877:711-712.

1983

Wood— Stork Phenetic Relationships

107

. 1878. On the trachea of Tantalus loculator and of Vanellus cayennensis. Proc.

Zool. Soc. London, 1878:625-629.

Harrison, C. J. O. (ed.). 1978. Bird families of the world. Harry N. Abrams, Inc.,
New York, 264 pp.

Hellack, J. J. 1976. Phenetic variation in the avian subfamily Cardinalinae. Occas.
Papers Mus. Nat. Hist., Univ. Kansas, 57:1-22.

Kahl, M. P. 1 966. Comparative ethology of the Ciconiidae. Part I. the Marabou Stork,
Leptoptilos crumeniferus (Lesson). Behaviour, 27:76-106.

. 1971a. Social behavior and taxonomic relationships of the storks. Living Bird,

10:151-170.

. \91\b. Food and feeding behavior of openbill storks. J. Omith., 1 12:21-35.

. 1972a. A revision of the family Ciconiidae (Aves). J. Zool., 167:451-461.

. \912b. Comparative ethology of the Ciconiidae. Part 2. The adjutant storks,

Leptoptilos dubius (Gmelin) and L. javanicus (Horsfield). Ardea, 60:97-1 1 1.

. 1 911c. Comparative ethology of the Ciconiidae. The wood-storks (genera Myc-

teria and Ibis). Ibis, 114:15-29.

. \911d. Comparative ethology of the Ciconiidae. Part 4. The “typical” storks

(genera Ciconia, Sphenorhynchus, Dissoura, and Euxenum). Z. TierpsychoL, 30:
225-252.

— . 1972^. Comparative ethology of the Ciconiidae. Part 5. The openbill storks

(genus Anastomus). J. Omith., 113:121-137.

. 1973. Comparative ethology of the Ciconiidae. Part 6. The Blacknecked, Sad-

dlebill, and Jabiru storks (genera Xenorhynchus, Ephippiorhynchus, and Jabiru).
Condor, 75:17-27.

. 1979a. Family Balaenicipitidae. Pp. 252-253, in Check-list of birds of the

world (E. Mayr and G. W. Cottrell, eds.), Mus. Comp. Zool., Cambridge, Massa-
chusetts, 2nd ed., Lxvii + 1-547.

. \919b. Family Ciconiidae. Pp. 245-252, in Check-list of birds of the world (E.

Mayr and G. W. Cottrell, eds.), Mus. Comp. Zool., Cambridge, Massachusetts, 2nd
ed., Lxvii + 1-547.

Katz, J . C. , and F. J . Rohlf. 1974. Functionplane — A new approach to simple structure
rotation. Psychometrika, 39:37-51.

Ligon, j. D. 1967. Relationships of the cathartid vultures. Occas. Papers Mus. Zool.,
Univ. Michigan, 651:1-26.

Mitchell, P. C. 1913. Observations on the anatomy of the Shoebill, Balaeniceps rex,
and allied birds. Proc. Zool. Soc. London, 1913:644-703.

Parker, W. K. 1 860. Abstract of notes on the osteology of Balaeniceps rex. Proc. Zool.
Soc. London, 1860:324-330.

Peters, J. L. 1931. Check-list of birds of the world. Harvard Univ. Press, Cambridge,
Massachusetts, l:xviii+ 1-345.

Robins, J. D., and G. D. Schnell. 1971. Skeletal analysis of the Ammodramus-
Ammospiza grassland sparrow complex: a numerical taxonomic study. Auk, 88:
567-590.

Rohlf, F. J. 1970. Linear adaptive hierarchical clustering schemes. Syst. Zool., 19:58-
82.

— — . 1981. Consensus indices for comparing classifications. Math. Biosci., 59:131-
144.

Rohlf, F. J., J. Kishpaugh, and D. Kirk. 1979. NT-SYS. Numerical taxonomy system
of multivariate statistical programs. SUNY, Stony Brook, New York.

Rohlf, F. J., and R. R. Sokal. 1980. Comments on taxonomic congruence. Syst.
Zool., 29:97-101.

. 1981. Comparing numerical taxonomic studies. Syst. Zool., 30:459-490.

Schnell, G. D. 1970. A phenetic study of the suborder Lari (Aves) 1. Methods and
results of principal components analyses. Syst. Zool., 19:35-57.

108

Annals of Carnegie Museum

VOL. 52

Sneath, P. H. a., and R. R. Sokal. 1973. Numerical taxonomy. W. H. Freeman &
Co., San Francisco, 573 pp.

Vanden Berge, J. C. 1970. A comparative study of the appendicular musculature of
the order Ciconiiformes. Amer. Midland Nat., 84:289-364.

Verheyen, R. 1959. Contribution a I’anatomie et a la systematique de base des Ci-
coniiformes (Parker, 1868). Bull. Inst. Roy. Soc. Nat. Belgium, 35:1-34.

Weldon, W. F. R. 1883. On some points in the anatomy of Phoenicopterus and its
allies. Proc. Zool. Soc. London, 1883:638-652.

Wetmore, a. 1 960. A classification for the birds of the world. Smithsonian Misc. Coll.,
139:1-37.

Wood, D. S. 1 979. Phenetic relationships within the family Gruidae. Wilson Bull., 9 1 :
384-399.

. 1982. A phenetic assessment of the Ciconiidae (Aves) using skeletal mor-
phology. Unpubl. Ph.D. diss., Univ. Oklahoma, Norman.

. 1983^. Character transformations in phenetic studies using continuous mor-
phometric variables. Syst. Zool., in press.

. 1983Z). Concordance between classifications of the Ciconiidae (Aves) based on

behavioral and morphologic data. Submitted to J. Ornith.

Appendix I

Specimens used in this study. The following abbreviations for museums are used:
AMNH, American Museum of Natural History, New York, New York; FM, Field Mu-
seum of Natural History, Chicago, Illinois; KU, Museum of Natural History, University
of Kansas, Lawrence, Kansas; MCZ, Museum of Comparative Zoology, Harvard Uni-
versity, Cambridge, Massachusetts; MVZ, Museum of Vertebrate Zoology, University
of California, Berkeley, California; NMNH, National Museum of Natural History,
Smithsonian Institution, Washington, D.C.; OU, Stovall Museum of Science and History,
University of Oklahoma, Norman, Oklahoma; UMMZ, Museum of Zoology, University
of Michigan, Ann Arbor, Michigan; YPM, Peabody Museum, Yale University, New
Haven, Connecticut

Mycteria americana- AMNW 605, 9856; KU 30867; NMNH 19532; OU 7720.
Mycteria NMNH 345229, 430169.

Mycteria zte- AMNH 2626; FM 1 04694; KU 7079 1 ; NMNH 431651; UMMZ 2 1 5038;
YPM 7526.

Mycteria leucocephala-^MHW 432506, 432507; UMMZ 153055, 154507; YPM 406.
Anastomus oscitans-NMNH 488759, 489353, 553234; UMMZ 216572, 216573.
Anastomus lamelligerus-MVZ 133407; NMNH 291418, 291419, 291420.

Ciconia nigra-MCZ 6991 \ NMNH 19784, 291560; YPM 4350.

Ciconia abdimii-AMNn 4860; NMNH 430455, 430456, 430528; YPM 7588.

Ciconia episcopus- AMNH 3553, 1370, 4952; NMNH 225807, 226001.

Ciconia mzz^ari-NMNH 19940; OU 10611; UMMZ 156986, 156987, 158608.
Ciconia ciconia- AMNH 1723, 1724, 2286; NMNH 343156, 430168; UMMZ 151110.
Ephippiorhynchus asiaticus— AMNH 1729, 2083, 3891; NMNH 19694, 346193.
Ephippiorhynchus senegalensis— AMNH 2433, 2903; FM 104380; UMMZ 211561,
211562.

Jabiru myc/rnTz-NMNU 343465; OU 6794, 10612, 14386; UMMZ 154788.
Leptoptilos 77zva«zcw5-AMNH 2721, 4392, 5059; NMNH 223897; UMMZ 218204.
Leptoptilos dubius- AMNH 4023; FM 104387; NMNH 225988, 429220; YPM 409.
Leptoptilos crumeniferus- AMNH 1731, 5862; NMNH 488129; OU 14385; UMMZ
210451, 218203.

Balaeniceps r^x-AMNH 5935, 8817; NMNH 344963, 345070; UMMZ 215884.
Scopus umbretta-NMNH 18898; UMMZ 154762, 158164, 185419, 158435.

1983

Wood— Stork Phenetic Relationships

109

Appendix II

Descriptions of the points, linear measurements and angular characters used in this
study. The letters identifying the points correspond to those labeled on Fig. 3. Nomen-
clature follows Baumel et al. (1979) and Bock and McEvey (1969). The linear measure-
ments are inter-point distances; they are listed as pairs of letters corresponding to the
described points. References to these measurements use a combination of the two letter
code for the bone and the two letters for the points (for example, SK-CR, TA-JP). The
interconnections between any three points form a triangle; angles subtended by the sides
of these triangles are used as angular characters. Specific angles are named using the
letters of the three points forming the triangle with the point at the vertex of the angle
listed second (that is, character AEB is the angle subtended by lines connecting point A
with point E and point B with point E). As with the linear measurements, references to
the characters include the code for the bone (for example, AX-BCH, HU-KMN). Since
all angles for each triangle are used as angular characters, only the triangles are listed.

Ossa cranii, ossa faciei— SK
Points

A. Os palatinum; lamella caudolateralis; caudolateralmost point

B. Os palatinum; crista ventralis; caudoventralmost point

C. Os palatinum; margo ventralis articularis pterygoideum; medialmost point

D. Lamina basisphenoidalis; processus medialis; rostralmost point

E. Os pterygoideum; facies articularis guadratum; medialmost point
E. Os quadratum; condylus medius; rostralmost point

G. Os quadratum; condylus medius; lateralmost point

H. Os quadratum; cotyla quadratojugalis; caudoventralmost point

I. Os quadratum; condylus caudalis; ventralmost point

L. Ala occipitalis; processus exoccipitalis; medioventralmost point

M. Foramen magnum; margo dorsalis; medial point

N. Processus suprameaticus; ventralmost point

O. Processus zygomaticus; rostralmost point

P. Processus postorbitalis; ventralmost point

Q. Processus orbitalis quadrati; medialmost point

R. Os parietalis; margo caudalis articularis supraoccipitalis; medialmost point
Linear measurements

AB AC BC CD CE CH CM CN CR DE DH DI DL DM DN EF

EG EH EN FG FH FI FM GH HI HL HN HO HQ IL IN LM

MR NP NR OP OQ OR PR

Triangles

ABC CDE CDH CDN CMR DLH DLI DLM EFG EGH ENH FHN FHI

HIN HOQ NPR OPR

Axis— AX
Points

A. Processus articularis caudalis; facies articularis; caudolateralmost point

B. Arcus axis; caudomedialmost point

C. Corpus axis; facies articularis caudalis; ventromedialmost point

D. Corpus axis; processus ventralis; ventralmost point

E. Corpus axis; facies articularis atlantica; ventralmost point

F. Processus articularis cranialis; facies articularis; cranialmost point

G. Arcus axis; craniomedialmost point

H. Area ligamentum elastici caudalis; dorsalmost point

110

Annals of Carnegie Museum

VOL. 52

Linear measurements

AB AC AD AE AF BC BD BE BG BH CD CE CG CH DE EF
EG FG

Triangles

ABC ABD AEF BCE BCG BCH CDE EFG

Humerus —HU

Points

A. Impressio m. pectoralis pars profundus; distalmost point

B. Crista pectoralis; cranialmost point

C. Impressio m. pectoralis pars profundus; proximalmost point

D. Tuberculum dorsale; dorsalmost point

E. Junctura intumescentia, margo ventralis

F. Impressio m. scapulohumeralis caudalis; distalmost point

G. Impressio m. biceps brachii; proximalmost point

H. Tuberculum ventrale; dorsolateralmost point

I. Impressio m. latissimus dorsi caudalis; proximalmost point

J. Impressio m. latissimus dorsi caudalis; distalmost point

K. Condylus dorsalis; cranioventralmost point

L. Impressio m. supinator; proximalmost point

M. Fossa m. brachialis; proximalmost point

N. Fossa m. brachialis; distalmost point

O. Tuberculum supracondylare ventrale

P. Epicondylus ventralis; ventralmost point

Q. Epicondylus dorsalis; caudodistalmost point

Linear measurements

AB AC AD AE BC BD BE BI CE DE DF DH DI DJ EF EH

FG FH FI GH IJ KL KM KN KO KP KQ LM LN LQ MN MO

NO NP OP OQ

Triangles

ABC ABE ACE ADE BDI BED DEF DEH DIF DIJ FGH KLM KLQ
KMN KMO KOP KOQ LMN NOP

Clavicula— FU
Points

A. Synostosis interclaviculare; point sinister to craniodorsalmost point; points A and
I are on smae plana paramediana

B. Processus acromialis; caudodorsalmost point; sinister

C. Processus acromialis; caudodorsalmost point; dexter

D. Apophysis furculae; facies articularis apex carinae sterna; dorsalmost point

E. Apophysis furculae; facies articularis apex carinae sterna; ventralmost point

G. Synostosis interclaviculare; craniodorsalmost point

H. Same as point G (in the Ciconiidae)

I. Apophysis furculae; facies articularis apex carinae sterna; lateralmost point

K. Point to give maximum thickness of extremitas stemalis claviculae at point A

L. Apophysis furculae; facies articularis apex carinae sterna; midpoint
Linear measurements

AB AC AD AE AG AH AI AK AL BC DE GH GL IK IL
Triangles

ABC ADE AGH AGL AIK AIL

1983

Wood— Stork Phenetic Relationships

111

Coracoideum — CO
Points

A. Angulus medialis; medialmost point

B. Facies articularis stemalis; lateralmost point

C. Processus lateralis; dorsolateralmost point

D. Facies articularis stemalis; crista ventralis; craniolateralmost point

E. Facies articularis stemalis; crista dorsalis; caudalmost point

F. Junctura facies articularis humeralis, cotyla scapularis; ventralmost point

H. Facies articularis humeralis; dorsalmost point

I. Impressio ligamentum acrocoracohumeralis; dorsolateralmost point

J. Processus procoracoideus; dorsalmost point

K. Facies articularis clavicularis; caudalmost point

L. Processus acrocoracoideus; tuberculum brachialis

Linear measurements

AB AC AD AJ BC BD BJ FH FI FJ HI HJ HL JK JL KX
Triangles

ABC ABD ABJ FHI FHJ JHL JLK

Sternum— ST

Points

A. Spina externa; cranialmost point

B. Spina interna sinister; dorsalmost point

C. Processus craniolateralis; craniolateralmost point

D. Tuberculum labri ventralis (junctura labrum ventralis, lineae intermusculare); sin-
ister

E. Apex carinae; facies articularis; ventralmost point

F. Margo caudalis; median point

G. Trabecula lateralis sinister; caudolateralmost point

H. Process costalis 2 sinister; facies articularis; midpoint

I. Facies articularis coracoideus dexter; medialmost point

Linear measurements

AB AC AD AE BC BD BE BH BI CD CE CF CH DE DF DG
DH DI EF EH FG

Triangles

ABD ABE ACD BCD BDE BDI BEH CDH DCF DFG ECF

Synsacrum, os coxae— SY
Points (all sinister or median)

A. Fovea costalis for caudalmost costa vertebralis

B. Fossa acetabuli; margo cranialis; cranialmost point

C. Antitrochanter; caudalmost point

D. Processus lateralis crista iliaca; lateralmost point

E. Foramen obturatum; margo caudalis; caudodorsalmost point

F. Processus terminalis ischii; caudalmost point

G. Foramen ilioischiadicum; caudalmost point

H. Ala preacetabularis ilii; lateralmost point

I. Extremitas caudalis synsacri; middorsal point

J. Processus dorsolateralis ilii; caudalmost point

K. Extremitas cranialis synsacri; ventromedialmost point

112

Annals of Carnegie Museum

VOL. 52

Linear measurements

AC AF AH AK BC BD BE CD CE CF CG Cl CJ CK DE EF
EG EJ FI HK IK

Triangles

ACF AHK BCE BDE CDE CEF CEG CEJ CFI CIK

Tibiotarsus— TI

Points

A. Pons supratendineus; margo proximalis

B. Pons supratendineus; margo distalis

C. Impressio lateralis ligamentum transversum; distalmost point

D. Tuberculum intercondylare; cranialmost point

E. Epicondylus medialis; medialmost point

F. Sulcus cartilaginis tibialis; proximomedialmost point

G. Impressio m. gastrocnemius medialis; distalmost point

H. Crista cnemialis cranialis; proximalmost point

I. Crista cnemialis lateralis; lateralmost point

J. Foramen interosseum distale; proximalmost point

K. Foramen interosseum proximale; distalmost point

L. Area interarticularis; proximalmost point

Linear measurements

AB AD BC BD BE BF CD EF EL GH GI GJ GK HI HJ HL
IK IL JK JL

Triangles

ABD BCD BEF GHI GIK GJK HLI HLJ

Tarsometatarsus— TA j

Points

A. Crista medialis hypotarsi; margo plantaris; proximalmost point |

B. Crista lateralis hypotarsi; margo plantaris; proximalmost point ;

C. Sulcus hypotarsi; facies plantaris; proximomedialmost point j

D. Eminentia intercondylaris; proximalmost point

E. Cotyla medialis; margo lateralis; plantolateralmost point ;

F. Cotyla medialis; margo dorsalis; dorsalmost point

G. Sulcus extensorius; midpoint between margo distalis impressiones retinaculi exten-

sorii I

H. Trochlea metatarsi III; facies plantaris; proximalmost point j|

I. Foramen vasculare distale; facies plantaris; margo distale i

J. Trochlea metatarsi IV; margo lateralis; proximalmost point

K. Trochlea metatarsi II; margo medialis; proximalmost point |;

L. Trochlea metatarsi IV; margo medialis; proximalmost point |

M. Trochlea metatarsi II; margo lateralis; proximalmost point |i

N. Trochlea metatarsi III; condylus lateralis; distalmost point i

P. Fossa metatarsi I; margo proximalis; proximalmost point

Linear measurements !

AB AC AE BC BE CD CE CF DF DG DN FG HI HJ HK HP j

IK IM IP JK JL JN JP KM KN LM LN LP MN

Triangles !

ABC ABE ACE CDF DFH HIK HIP HJK HJP IMK JKN JLP LMN I

A'

)

■%

\

'h

ISSN 0097-4463

■7 73

AN NALS

of CARNEGIE .MUSEUM

CARNEGIE MUSEUM OF NATURAL HISTORY

4400 FORBES AVENUE « PITTSBURGH, PENNSYLVANIA 15213
VOLUME 52 17 JUNE 1983 ARTICLE 6

SYSTEMATICS OF LIOPHIS REGINAE AND
L. WILLIAMSI (SERPENTES, COLUBRIDAE),

WITH A DESCRIPTION OF A NEW SPECIES

James R. Dixon*

Abstract

The currently recognized species Liophis zweifeli is reduced to a subspecies of L. reginae
and L. oligolepis is shown to be a junior synonym of L. reginae semilineata. Geographic
variation in L. reginae is discussed in relation to major physiographic zones and the
relationship of L. reginae to L. epinephelus is briefly discussed. Liophis williamsi is
redefined, but its relationship to L. reginae is unclear. Liophis andinus, sp. nov., is
described from the Cochabamba area, Bolivia.

Introduction

The species discussed herein have cis-Andean South American dis-
tributions. Liophis williamsi and a new species described herein occur
in widely separated, middle elevation Andean locations. Whereas L.
reginae also occurs at middle elevations, it is a species primarily of
lowlands, from Colombia to Argentina. Liophis reginae has the widest
distribution of any South American Liophis.

Liophis reginae has seven names associated with it that are princi-
pally based on variant color patterns. The descriptions of violaceus
(Lacepede, 11 S9), graphicus {Shaw, 1802), semilineata {WaglQv, 1824),
reginae macrostoma {Amaral, 1935), reginae maculicauda {HogQ, 1954),
and zweifeli (Roze, 1959) are based on color morphs, whereas reginae

* Address: Department of Wildlife and Fisheries Sciences, Texas A&M University, Col-
lege Station, TX 77843.

Submitted 22 October 1982. 7/

113

114

Annals of Carnegie Museum

VOL. 52

l:

(Linnaeus, 1758) and oligolepis (Boulenger, 1905) are based on both I
quantitative and color pattern characters. Liophis williamsi has a single |
color pattern morph that is repeated in one population of L. epinephelus Si
from Peru. Liophis andinus exhibits color pattern variation similar to !
that of L. reginae. ||

Liophis reginae is closely related to L. epinephelus. The latter species j!
is discussed elsewhere (Dixon, 1983), but for the purpose of comparison ||
with reginae, the following points obtain: the two species are similar !
in most characters in some parts of their ranges (Fig. lA, B, C, and |
Table 1 ), but they are primarily allopatric, epinephelus being, in general. Si
trans- Andean and reginae cis- Andean. The two species are parapatric
along the eastern slope of the Ecuadoran Andes, principally in the upper ii
Rio Pastaza and the upper Rio Santiago basins at about 1 800 m. Where ::
they are parapatric, their color patterns are not identical. Liophis regi- |
nae has a posterior lateral black stripe that occurs as a narrow border |
between scale rows three and four, whereas in L. epinephelus the stripe !l
covers most of the third, from one-third to all of the fourth, and oc- j
casionally part of the fifth scale row. ij

|l

Systematic Account

Liophis reginae (Linnaeus) I

Distribution.— Liophis reginae is restricted to cis-Andean South j
America, where it occurs in every country except Chile and Uruguay, i
Its distribution extends from southern Brazil and northern Argentina |
to Trinidad and Venezuela (Fig. 2). I

Lectotype. — ThQ original description of Coluber reginae by Linnaeus !
(1758) gave the number of ventrals as 137 and subcaudals as 70. Ac- i
cording to Andersson ( 1 899), the Drottingholmense Museum (Museum I
Regis Adolphi Friderici I) has a jar labeled C. reginae that contains
two young specimens of L. reginae (ventrals 137, 141; subcaudals 76, |
81, respectively) and one young specimen of Liophis lineatus. An- I
dersson (1899) thought that the latter specimen had been put among
the young L. reginae by mistake because at first glance they appear
similar. Andersson suggested that these were the specimens upon which I
Linnaeus based his description and by inference, the specimen with ||
137 ventrals should be the lectotype. Based upon the number of ventrals j
and subcaudals given by Andersson, there are two possible geographical In
areas from which the specimens may have come: the central coast zone U
of Brazil or the Suriname/Guyana area. The latter region is more likely ;
because many of Linnaeus’ South American specimens came from |i
Suriname through the various directors of the East India Company
(Holm, 1957). I accept Andersson’s (1899) lectotype designation, and jj
suggest the restriction of the type-locality to ‘‘Suriname.”

Table \.— Variation in Liophis reginae and L. epinephelus for ventrals, caudals, and tail/total length ratios. Am. = middle and western
Amazon, exclusive of Perd; Para = eastern Amazon; Mt. ^ Mato Grosso; NE = northeastern states; RJ = Rio de Janeiro region; SP =
Sdo Paulo region; Par/Argen. == Paraguay /Argentina sample. N= number in sample; M= mean; SE = standard error.

1983

Dixon— System ATics of Liophis

115

o

OS

(N

<N

m

^ 1

q

<N

m

fN|

<N

q

d

d

d

d

d

d

d

d

o '

d

d

d

d

d

d

d

d

r--

o\

(N

q

q

m

q

r-

q

q

r-

od

r-’

od

od

'T

SO

ITS

iri

d 1

so

od

so

os

rd

<N

rd

(N

rM

<N

<N

fN

(N

(N

(N

(N

<N

(N

<N

(N

<N

(N

fs|

(N

(N

m

O

q

00

m

r-

oo

q

r-

r-

r-

q

OS

q

q

q

d

d

d

r-’

od

od

os

so

od

os I

od

d

od

«/d

d

so

so

ro

1

1

ro

1

1

<N

1

(N

1

(N

1

(N

1

fN

1

(N

1

(N '

1 1

<N

1

C<S

1

<N

1

<N

1

<N

1

<N

1

rN

1

<N

1

1

1

1

1

1

m

1

r-

1

<N

1

(N

1

q

1

q

1 1

1

q

1

q

1

q

1

q

1

1

q

1

1

q

r--'

ir>

rd

d

d

cd

d

q I

rd

r--’

fd

od

r--"

os

Os

od

<N

<N

(N

(N

CM

(N

<N

(N

(N

fN

(N

<N

(N

(N

•— H

<N

On

>r>

r-

OS

fN O

oo

r-

<N

O

OM

so

00^00 04

q

fTS

q Os

Os so

q

oo

d '-<■ d

<N

d

d

-4 d

d ^4 rd —I -d

d

d

d

d

d

Os

q

q

Os

q

q

q

q

q

q

m

m

<N

OS

q

rN

q

m

q

<N

d

os

od

od

id

d

od

d

od

(N

d

so

q

d

so

r-

r-

oo

so

so

so

so

so

so

OO

oo

l/s

IT)

so

so

so

oo

o

rrs

oo

so

r-

oo

O

so

OS

os

rsi

r-

(N

o

»rs

r-

1

oo

1

oo

00

1

1

1

oo

1

r-

1

1

oo

1

OO

1

00

1

OO

1

1

so

1

ir>

1

r"

1

1

1

1

m

!

oo

i

1

Os

1

i

so

1

IT)

1

o

1

(N

1

r-

1

o

1

1

m

I

ITS

1

r-

1

I

m

1

os

!

<N

i

so

so

m

so

SO

m

VO

VO

r-

'TS

'T

ITS

-H ro ro

«rs OO

so

— 1 ^ ^ rr> OS

ITS

SO

o

m

m

<N (N

^ (N

m <N ^ — 1 -H

o

oo

r-

oo

r-

q q

q

q

q q

so

00

os

IT) IT)

oo

so

IT) m

d

d

d

d -4

d

d

-4 d

d

d

d

d d

d

d

d

d

d d

os

r--

q

m

in

Os

Os

OO

q

q

oo

q

q

q

q

r-

oo

(4

od

(N

(N

d

so

d

SO

OS

od

c6

(N

(N

d

so

od

d

rn

m

m

m

in

m

in

in

"—I

I— <

>—1

<— 1

y—l

.—1

<— 1

— H

•— 1

<— <

. — 1

>—1

»— 1

■-H

.— I

r-

SO

oo

o

in

m

m

o

m

o

oo

os

CM

oo

<N

m

'Tt

in

m

m

m

so

m

m

m

in

m

m

m

so

so

q

q

q

q

qi

q

q

q

q

q

q

q

q

q

q

q

q

q

1

1

os

1

r-

1

rs

1

r-

m

1

r-

1

oo

1

oo

1

oo

in

1

m

1

so

1

o

1

o

1

os

1

in

1

oo

1

<N

<n

m

m

m

m

m

Ti-

m

in

m

m

m

CN

<r-H

•— 1

•— 1

•—1

f-H

•— <

<— 1

"—I

>— <

>—1

•— <

<—1

•—1

<— 1

I— 1

1

r-

rsi (N

— ■ oo os

r-

so

in

o^ ^ ^ as o

^ o

m

m

<N

m — 1

m r--

(N

<N

m

(N ^ 1-H ^

in in

o

os

(N

m

d

.Co p

^ I '1^

s;

lamonae 48 141-157 149.7 0.7 41 51-67 59.2 0.7 22 20.2-24.1 22.2

opisthotaenia 32 142-152 147.0 0.5 29 53-78 64.7 0.9 17 20.2-25.6 23.6

bimaculatus 41 162-191 174.9 1.0 34 44-80 63.2 1.0 17 18.9-24.0 21.3

116

Annals of Carnegie Museum

VOL. 52

Fig. 1. — Comparison of Liophis reginae and L. epinephalus in: A) numbers of maxillary
teeth.

Variation in color pattern.— ThQ color pattern of 80% of the speci-
mens examined of L. reginae consists of an olive to grayish green
dorsum, with faint to well-marked dorsolateral diagonal black streaks
or spots on the anterior one-third of the body. A thin black lateral line
begins on the border of scale rows three and four on the posterior one-
fourth of the body and continues to the tip of the tail. Occasionally,
this line may be present only above the vent or slightly in front of it

1983

Dixon— System ATics of Liophis

117

Fig. 1 (cont.)— Comparison of Liophis reginae and L. epinephalus in: B) comparison of
the site of the reduction from 17 scale rows to 15; C) range and mean numbers (in
parentheses, followed by sample size) of ventrals at various elevations along the Pacific
and Atlantic versants of the Ecuadorean Andes (single set of numbers, in brackets on
the right in 1C, represent the only population of L. epinephelus that appears to be
parapatric with L. reginae).

118

Annals of Carnegie Museum

VOL. 52

Fig. 2.— The distribution of Liophis reginae, L. epinephelus, L. williamsi, and L. andinus
in South America. Circular dots represent the distribution of L. reginae, black stars
represent epinephelus, and the square represents L. andinus. The insert shows the dis-
tribution of L. williamsi (open star) and sympatric localities of williamsi and reginae

1983

Dixon— System ATics of Liophis

119

thence continuing to the tip of the tail. The venter is checkered yellow
and black in 95% of individuals and the subcaudal area is white. The
ventral surface of some females in some populations is immaculate
white (rare in males).

A sample of 100 individuals of L. reginae from Iquitos, Peru, con-
tains at least three discrete patterns. One pattern is distinctly “salt and
pepper” with each scale reddish anteriorly, greenish medially, and black
posteriorly. Another pattern has the first and second scale rows pale
green or yellowish green. The anterior part of the body is bright red
and the posterior part dull green, with a black lateral stripe on the
borders of the third and fourth scale rows continuing to the tip of the
tail. The third pattern is the one previously described for about 80%
of individuals. A combined form, with elements of all three patterns,
can occasionally be found in a single individual.

The Peruvian sample seems to have the highest number (18%) of
individuals with immaculate venters. Three hundred and two of 395
individuals have 17-17-15 scale rows and checkered venters. Sixty-
nine males from Peru with 17-17-15 scale rows have checkered ven-
ters—none were immaculate white. The holotype of L. oligolepis Bou-
lenger (Fig. 3D) is the only available male with the combination of
15-15-15 scale rows and a white venter. Of 107 Peruvian females with
17-17-15 scale rows, four have white venters and 103 are checkered;
of 53 Peruvian females with 15-15-15 scale rows, 35 have white venters
and 1 5 are checkered. These data suggest that scale row reductions and
ventral color in Peruvian reginae is sex-linked in females, and only
rarely so in males.

The subcaudal area is usually white, yellow, or salmon. However,
specimens from the population in central Brazil and northwestern Ar-
gentina tend to have a few black spots on the outer edges of the sub-
caudals. In preservative, these spots appear as black smudges.

Geographic association of patterns seems to be discordant on the
north and west sides of the Amazon Basin. The “salt and pepper”
pattern is common in the coastal Andes of Venezuela (Fig. 3C), part
of the Tepui region of the Guiana Shield, Trinidad, and extreme north-
ern Guyana. However, this same pattern is present in the eastern middle
slopes of the Andes of central Peru and continues southward into
Bolivia. The pattern also shows up as an occasional variant in the mid-
Amazon Basin, Iquitos area of Peru, and extreme western Venezuela.

(circle/white star) and localities of reginae (black circles). The heavy black lines denote
the approximate boundaries of the subspecies of L. reginae (Z = zweifeli; R = reginae;
S == semilineata: M = macrostoma). Dashed lines around circular dots represents zones
of intergradation.

120

Annals of Carnegie Museum

VOL. 52

D

Fig. 3. —A) paratype of Liophis williamsi (MBUCV 3044), Rancho Grande, Venezuela. i
B)Holotype of Liophis graphicus (BMNH 1946.1.5.73), “America, ’’represents/., reginae i
reginae. C) Holotype of Liophis zweifeli (MBUCV 95), Rancho Grande, Venezuela, ij
represents L. reginae zweifeli. D) Holotype of Liophis oligolepis (BMNH 1946.1.4.66),
Igape-Assu, Brazil, represents L. reginae semilineata.

The light-side, brightly colored pattern (Fig. 3B) is frequently seen in
the lowlands of central and southeastern Peru, but more commonly in
the Guianas (southern Guyana, Suriname, and French Guiana). The

1983

Dixon— System ATics of Liophis

121

latter pattern is supplemented with dorsolateral light flecking and/or
dark spotting in Suriname and southern part of Guyana.

The typical pattern of an olive green dorsum with faint dorsolateral
black streaks anteriorly and a thin, black lateral line posteriorly is
prevalent throughout Amazonian South America in lowland situations,
including the Amazon tributaries that extend into central portions of
Brazil. The same pattern, but with a paler hue, is present in the Parana
drainage of Paraguay. Specimens from the Brazilian states of Sao Paulo
and Parana tend to have denser dorsolateral spotting anteriorly, bright-
er colors, and the subcaudals more often smudged with black.

A single female specimen of L. reginae from Tobago (USNM 228069)
has an atypical color pattern. It resembles the pattern of L. epinephelus
lamonae {sensu stricto, Dixon, 1983) from the middle elevations of
eastern Colombia. The dorsal pattern (at 100th ventral) consists of the
first and second scale rows olive green; third scale row olive green with
a large black smudge in the middle of the scale; fourth scale row tan
(cream in life?) with the lower edge black; fifth scale row tan with the
upper one-third with greenish black flecks; sixth to tenth scale rows
olive green. The ventrals and subcaudals are immaculate cream. The
absence of black marks on the ventrals that are typical of most L.
reginae, may be sex-linked as indicated for females with immaculate
ventrals from the Amazon Basin (this paper). The dorsolateral tan stripe
is the most striking difference between the patterns of the specimens
from Tobago and L. reginae from Trinidad (the latter lack this stripe).

Other characters of the Tobago specimen follow: 27 maxillary teeth,
17-17-15 scale rows, 143 ventrals, 77 subcaudals, 8-8 subcaudals, 8-8
supralabials, 9-10 infralabials, 4+5 supralabials entering orbit, 1+2
temporals, 1-1 preoculars, 2-2 postoculars, divided anal plate, 482 mm
in total length, 120 mm tail length; 0.270 tail/total length ratio. All of
the latter characters are typical for most samples of mainland and
Trinidadian L. reginae (see Tables 1 and 2). Hardy (1982) has com-
mented on this specimen and presented a photograph.

Scutellation variation. —The number of head scales varies little be-
tween and among samples of L. reginae from throughout its geographic
range. The distribution of the numbers of head scales (Table 2) of L.
reginae is typical for most species of Liophis.

A comparison of the usually sexually dimorphic characters (ventrals
and caudals) between males and females of L. reginae from Amazonian
Peru (sample of 67 5(5 and 174 29) reveals no significant differences in
mean numbers. Therefore, sexes were combined for analyses of these
characters in sample by sample comparisons. An examination of the
average number of ventrals throughout the cis-Andean distribution of
L. reginae (Fig. 4) shows that the Amazonian population (samples 7-

122

Annals of Carnegie Museum

VOL. 52

SOUTH AMERICA

Fig. 4.— A) Geographic locations of samples of L. reginae, arranged by latitude from
North to South, and by longitude from East to West.

1983

Dixon— System ATics of Liophis

123

Fig. 4. — B) Univariate analysis of ventrals and subcaudals arranged in order of the samples Fig. 4 A. Horizontal line represents the
range of variation, vertical line the mean, black rectangle two standard errors on either side of the mean, open rectangle one standard
deviation on either side of the mean.

124

Annals of Carnegie Museum

VOL. 52

.N

So

G

o

K

>3

so

II

^ ^

(N

cn

(N

^ ^

o^

o

'O

2

Os

o

4+5

+

cn

+

^ V

(N

.s;

so

m

o\

o

ir3|

VO

1

oo

+

'd'l

+

ai

oo

^ ^

^ ^

c3

'O

(N

.—1

C

Tj-

s-.^

’Si)

so

oo

Os

+

VO

in

oo

Os

IT)

+

+

>o

a

_o

r— t

.-M

m

?-iH

' — '

' — ^

'W'

s;

m

o^

so

rs)

rs)

r-

+

+

-i

OO

sn

+

rs)

Co

<N

cn

rs)

G

00

^ ^

^ ^

so

VO

m

rs)

O'

Co

<N

(N

oo

o

rs)

rN

1

<N

1

+

+

X.

O

oo

-o

;s

oo

iro

cn

.s;

<N

<N

r-

os

|cr)

— '|rs)

^ 1

1

1

1

OO

+

+

+ +

s;

m

1'^

.o

o

^ ^

s;

t--

«

r-

<N

.— (

m

rs)

irv

w

' — ^

c ^

s ^

' — '

s ^

1

.—1

r-

oo

cn

.—1

(N

J.

-

so

00

+

lO

+

+

+

3

m

cd

H

in

in

in

in

3

in

3

in

Uh

23

3

3

o

3

3

3

3

cd

3

3

■-e

o

2

o

o

o

3

3

Vh

cx

o

u

3

Vh

c

(D

c/3

CX

cjo

a

c

U

o

:3

fH

o

' a

a>

Oh

oo

oo W

H

I

1983

Dixon— Systematics of Liophis

125

1 9) is intermediate, with ventral numbers increasing to the south and
decreasing to the northeast (samples 3-6). The greatest difference be-
tween adjacent samples is between Suriname and eastern Amazonia.
I have not examined, however, specimens from the area between Be-
lem, Brazil (sample 9a), and French Guiana (sample 4). There may be
a cline in ventral number between these two regions, or perhaps partial
gene flow around the Guiana Shield. There do not appear to be well-
defined steps in the ventral counts of the cis-Andean populations, ex-
cept the Venezuelan-Trinidad versus the Guyana-Surinam samples,
and the coastal Brazil sample (sample 31), and its adjacent, but allpatric
samples (samples 24a, 26a, 27, 28, 29a) from inland southeast Brazil.
The latter discordance is observed in several other species of snakes
with similar distributions {Typhlops brongersmianus, Chironius fuscus,
Liophis miliaris, Liophis poecilo gyrus). Geological and other historical
evidence suggests that the vegetation and climate of that particular
coastal zone of Brazil was probably connected to the Amazonian Forest
within the past few thousand years, and gene flow between populations
of both areas was relatively recent or perhaps continuous to the present
time (Dixon, 1979).

The number of subcaudals also shows both concordant and discor-
dant variation patterns (Fig. 4). Liophis reginae samples from northern
Venezuela, Trinidad, and eastern Venezuela (samples 1, 2, 7) have high
numbers of subcaudals. Adjacent Guyana, Suriname, and Colombia
snakes (samples 3, 4, 5, 6, 12b) have low numbers of subcaudals, as
do those from the Atlantic coastal zone of Brazil and most of western
cis-Andean South America. Paraguay and eastern Brazil (exclusive of
Atlantic Coast) samples have high numbers of subcaudals. Samples of
L. reginae from south of the Amazon Basin suggest that there is little
gene flow as expressed in subcaudal numbers, between the caatinga/
cerrado populations of Brazil and the southern Peru/eastern Bolivia
populations along any given latitude between 1 0° to 20°S. Samples from
near Isla Bananal, Brazil (19a) and northern Argentina (29b) tend to
be intermediate between western South American samples (18b, 18c,
19b, 20, 21b, 22, 23b, 24b, 26b) and eastern South American samples
(18a, 21a, 23a, 24a, 25a, 26a, 27, 28, 29a, 30). The difference observed
may be an artifact of combined samples along the various latitudes or
they may reflect true differences between allopatric samples. In either
case, the result appears as a “rassenkreis” (Fig. 4) because adequate
samples are not available from Pantanal region of western Brazil, east-
ern Bolivia, and northern Paraguay. The tail length/total length ratios
follow the same pattern as the subcaudals, but the numbers of maxillary
teeth are totally discordant.

Two hundred twenty seven L. reginae from the Peruvian Amazon
show a relatively smooth, north to south, cline in most characters

126

Annals of Carnegie Museum

VOL. 52 j:

examined. The number of ventrals, subcaudals, and tail length/total
length ratios increase (11b, 12c, 14, 15, 16, 17b, 18c, 22), whereas the
number of maxillary teeth decreases. A comparison of large samples j
(Ecuador versus Peru) suggests that some characters of the populations |
between these two countries are signihcantly different. However, when
subsets of the adjacent samples (northern Peru, southern Ecuador) are !
compared, there is a relatively smooth dine between the samples. This !
merely suggests that some of the significant differences between samples j

are the result of clustering of subsets of samples by country, and not
by geographic zones. The entire Amazon sample shows little variation
from east to west along the same latitude and vegetation zone (9a, 10a, i
11a, 11b, 12a, 12b, 12c). However, considerable north to south vari- [
ation is noted along different latitudes.

Allocation of names. — ThQXQ are seven names associated with L.
reginae (the names listed below are those recognized as subspecies, and [
synonyms):

reginae reginae (Linnaeus, 1758; Fig. 3B) i

(Lacepede, 1789) J

graphicus (Shaw, 1802)

Guyana, Suriname, and French Guiana !

reginae semilineata (Wagler, 1824; Fig. 3D)

6>%o/c/?A(Boulenger, 1905)

Amazonian Ecuador, Colombia, Peru, Bolivia, Venezuela, Brazil, also ;
Atlantic Forest of Brazil I

reginae macrostoma (Amaral, 1935) |

reginae maculicauda (Hoge, 1954)

Chaco Boreal and Cerrado vegetation of Brazil, Argentina, Paraguay,
and Bolivia

reginae zweifeli (Roze, 1959; Fig. 3C) i

Montane zones of Venezuela and Trinidad

Three names, ornata (Jan, 1863), viridicyanea (Jan and Sordelli, |
1866), and maculata (Steindachner, 1867), assumed to be varieties of jl
L. reginae, belong to other species. The holotype of ornata (MHNG
108:39) was examined and found to be an example of L. miliaris. The
holotype of viridicynaea has not been located, but the color pattern,
squamation and illustration furnished by Jan and Sordelli (1866) in-
dicate that this name belongs to L. poecilogyrus. The syntypic material
of maculata is apparently lost. Franz Tiedemann of the Austrian Nat-
ural History Museum (personal communication) stated that they have i

1983

Dixon— System ATics of Liophis

127

no Natterer material from “Ypenema,” Brazil, the type locality, but
have 32 specimens of L. reginae collected by Natterer from other
Brazilian localities. A detailed examination of Steindachner’s (1867)
description of maculata suggests that he had more than one species
present, and possibly two genera. He speaks of scale rows around the
body and subcaudal number as 1 5 and 62-68 in young individuals and
1 7 and 80-90 in old specimens, respectively, and of highly variable
color patterns. The squamation and color descriptions presented could
represent Brazilian Rhadinaea as well as Liophis. Until Steindachner’s
series of specimens are located, his name must reside as species in-
quir endue.

Key to subspecies. — The following key is provided to aid in recog-
nizing subspecies of L. reginae. The characters utilized are those that
are average for those populations occupying specific geographic areas
(Fig. 2).

1 . Dorsum spotted with black and yellow; black lateral caudal stripe faint or absent .
2

- Dorsum greenish, olive, grayish, never yellow and black spotted; black lateral caudal

stripe always present and distinct 3

2. Subcaudals average 80 (69-88), (Venezuela, Trinidad, northern Guyana) . . zweifeli

- Subcaudals average 65 (55-78), (cis-Andean middle slopes of Peru and Bolivia) . .
semilineata

3. Scale rows one and two pale colored 4

- Scale rows one and two colored like rest of dorsum 5

4. Dorsum with dense pale and dark paravertebral flecking, subcaudals average 74 (63-

80), (French Guiana, Suriname, Guyana) reginae

- Dorsum without dense pale and dark paravertebral flecking, subcaudals average 67

(57-71), (Amazonian lowlands of central and southern Peru) semilineata

5. Subcaudal area with ventrolateral black spots, flecks, or smudges, subcaudals average

81 (75-91), (southern Goias, Sao Paulo and south and southwest Parana, Brazil and
Argentina) macrostoma

- Subcaudal area immaculate, subcaudals average 70 (55-81), (Amazonian South

America and Atlantic coastal Brazil) semilineata

Liophis williamsi (Roze)

Fig. 3A

Distribution. —Liophis williamsi is known only from the ground-floor
leaf litter of cloud forest zones of the eastern Andes of Venezuela, from
Rancho Grande, Aragua, to Cerro El Avila, Distrito Federal (Fig. 2).

Type series. — Roze' s (1958) name is the only one that has been
applied to this species. Marcuzzi (1 950) mistakenly discussed this species
under Leimadophis bimaculatus opisthotaenia (=L. epinephelus), a
species that occurs in the western part of the Venezuelan Andes near
Merida.

I have examined the type series and four additional specimens. There
is strong sexual dimorphism in the number of ventrals and subcaudals.

128

Annals of Carnegie Museum

VOL. 52

but not in tail length/total length ratios, maxillary teeth, or color pat-
tern.

Variation.— Th.Q ventrals of four males vary from 155-169 (mean ^
1 60.3); subcaudals vary from 58-66 (mean 60.8). Three females have
ventrals that vary from 146-150 (mean == 147.4); subcaudals 53-62
(mean = 58.0). The maxillary teeth vary from 19-22 (mean = 20.5).
The tail length/total length ratios (%) vary from 20.7-22.6 (mean =
21.7). There are no scale row reductions, all specimens having 17-17-
1 7 dorsal scale rows. Supralabials, supralabials entering orbit, loreals,
and postoculars are invariably 7, 3+4, 1, 2, respectively. The anal plate
is divided. The infralabials vary as follows: 9-9 (4), 10-10 (2); preoculars
0-1 (1), 1-1 (4), 2-2 (2); temporals 1+2 (6), 1 + 1 (1). Maximum total
length is 465 mm in males and females.

Pattern.— The dorsal color pattern consists of anterior blackish spots,
bars and/or blotches surrounded by a dark brown ground color. At
midbody the darker spots tend to form black paravertebral lines on
scale rows seven or seven and eight. The zone between the fourth and
seventh scale rows is pale brown or yellowish brown, the upper half of
the fourth scale row is pale yellow, lower half of fourth and upper two-
thirds of third black, lower one-third of third yellowish, and all of
second and first scale rows brown. The dorsolateral dark stripes unite
above the tail and form a single line to the tip of the tail. The dorsal
surface of the head varies from brown to blackish brown, with or
without obscure black spotting. A black line extends from below the
nostril to the last supralabial and occasionally onto the nape, usually
covering the upper one-fourth of the supralabials. The middle portion
of the supralabials is white, whereas the lower one-third may be entirely
black, or white with large black spots concentrated on the central por-
tion of the lower one-third of each labial. The white area of the su-
pralabial may extend past the last supralabial to the third or fourth
scale row of the nape. The infralabials may be entirely black, or mostly
black with white edging. The chin and throat may be white with a few
black marks, or almost entirely black with a few white-edged scales.
The venter and subcaudal surface may be completely white with a few
black flecks, or venter densely spotted with black on a gray to pinkish
ground color with the subcaudals gray, with or without a midventral
black line extending from anal plate to tip of the tail.

Hemipenis.— The hemipenis has a smooth apical disc, and a sinistral
sulcus that forks one-third to one-half its distance from the-'base. The
structure is very spinose, without calyces. Roze (1958) indicated that
the specimen he examined had a calyculate zone near the apex of the
sulcus spermaticus, but neither Myers (1974:20) nor I have been able
to locate such a zone on the specimens we have examined.

1983

Dixon— Systematics of Liophis

129

Comments.— A color pattern similar to that of the head and trunk
of L. williamsi is also present in L. epinephelus fraseri from central
Peru. The patterns are so similar that I originally thought L. williamsi
was represented by two widely disjunct allopatric populations. However,
after examining a large number of individuals of the L. reginae group,
I am convinced that pattern types within this complex are repetitive
within and between species, and pattern alone will not insure proper
identification.

One additional species seems related to this complex, an altitudinal
isolate from Cochabamba, Bolivia, that shows a scale row reduction
mode similar to that found in some individuals of L. reginae. The
general shape of the head and body, the tail/total length ratios, and
variable color pattern suggest a relationship to both L. reginae and L.
epinephelus. I designate this undescribed population as:

Liophis andinus new species
Fig. 5 and 6

Holotype. —CamegiQ Museum of Natural History (CM) R2808, adult
male, from Incachaca, 2500 m, Cochabamba, Bolivia, collected by J.
Steinbach in October 1921.

Paratopotypes. -CM R2777, R2797-98, R2780-81, R2804-07;
American Museum of Natural History 36012, 36014, 36016.

Distribution. — Kjiown only from the type-locality (Fig. 2).

Diagnosis. —Liophis andinus is distinguished from all species of Lio-
phis except L. reginae by having 15-15-15 dorsal scale rows, generally
without reduction (one reduced to 14 posteriorly), rather than 19-17-
17, 19-17-15, 17-17-17, or 17-17-15 dorsal scale rows; and from all
species of Liophis by having three (very rarely two) supralabials entering
the orbit rather than two.

Description of holotype.— Adult male, total length 485 mm; tail length
1 30 mm; tail/total length ratio 0.268; head length 15.8 mm, head width
8.0 mm; diameter of orbit 2.7 mm; nostril to eye distance 2.8 mm;
ventrals 150, subcaudals 68; supralabials and infralabials 8-8; preoc-
ulars and postoculars 2-2; temporals 1+2; loreal 1-1; third, fourth, and
fifth supralabial entering orbit on each side; maxillary teeth 24, last
two enlarged and separated from remainder by a diastema equal in
width to the basal length of three prediastemal teeth; in situ hemipenis
9 subcaudals in length, slightly bilobed, forked sulcus spermaticus,
naked basal pocket, smooth distal, apical disc. Color pattern (in pre-
servative) as follows: top of head brown or olive brown, side of head
with broad black streak from edge of rostral, through eye to above and/
or slightly beyond last supralabial; upper edge of preocular with small
cream spot; infralabials, throat and first few ventrals immaculate cream;

130

Annals of Carnegie Museum

VOL. 52

I cm

Fig. 5.— A dorsal, ventral, and lateral view of the head of the holotype of Liophis andinus
(CM R2808).

supralabials cream except for dorsal edge which is black; midbody
pattern consists of cream ventrals with dark scattered flecks; upper edge
of first scale row and lower edge of second scale row with small black
dot at their alternate apexes, a thin black edge on each of these two
scale rows connects the small black dots, giving an appearance of a zig-
zag line; upper half of third and lower third of fourth scale rows with
straight-edged black line; upper third of sixth and all of seventh scale
rows black, lower edge straight, upper edge undulating; eighth (mid-
dorsal) scale row faintly speckled, otherwise cream; lateral interspaces

1983

Dixon— System ATics of Liophis

131

Fig. 6.— The midbody color patterns of the type series of Liophis andinus with the
exception of CM R2797 which appears to be patternless. A) represents CM R2798; B)
CM R2780; C) CM R2781, R2804, R2806-07, AMNH 36012, 36014, 36016; D) CM
R2777, R2805, R2808.

132

Annals of Carnegie Museum

VOL. 52

between black lines probably cream, but specimen is darkened in for-
malin and light ground colors are obscure; black stripe of third and
fourth scale rows continues onto the tail where it occupies the upper
half of the first and lower tip of the proximal lateral caudal scales, and
passes through the middle third of the first scale row distally; subcaudals
immaculate cream.

Variation.— T\iQ following data are based upon the type series; means
and sample size are in parentheses, measurements in millimeters; total
length of five males with complete tails 406-570 (490.0), seven females
445-700 (606.6); tail/total length ratios of males 0.268-0.283 (0.272),
females 0.238-0.261 (0.253); head length of both sexes 14.4-21.6 (17.8);
head width 8.0-11.7 (10.0); nostril to eye distance 2. 8-4.9 (4.3); di-
ameter of orbit 2. 7-4. 9 (4.28); head width/head length ratios 0.51-
0.59 (0.543); ventrals of males 148-151 (149.8), females 153-156
(154.6); subcaudals of males 67-72 (70.1), females 64-68 (65.8); max-
illary teeth of males 23-26 (24.3) females, 24-27 (25.0). Scale features
of the head vary as follows: preoculars 1-1 (8), 1-2 (1), 2-2 (4); post-
oculars 2-2 in all; loreal 1-1 in all; supralabials 8-8 (11), 7-8 (1), 8-9
(1); infralabials, 8-8 (8), 7-7 (1), 7-8 (1), 8-9 (1), 9-9 (2); temporals,
1+2 (11), 1+3 (2); supralabials entering orbit, 3+4+5 (11), 3+4+5/
4+5+6 (1), 3+473+4+5 (1).

The color pattern is obscure in most specimens because of darkening
by formalin. However, “in fluid” observation of the pattern reveals at
least five midbody patterns (Fig. 6, unicolored pattern not figured), of
which figures C and D represent about 83% of the individuals. The
venter appears to be cream or yellow with some dark flecking in all
individuals. The subcaudal area appears to be immaculate cream or
yellow. The cream or yellow spot on the upper edge of the preocular
is distinct in 10 individuals and obscure in three. In one specimen, an
obscure cream spot also appears on the upper edge of the suprapost-
ocular.

The in situ hemipenis varies in length from 8 to 9^2 (8.5) subcaudals.
It is slightly bilobed, the lobes usually indicated at the level of the
seventh or eighth subcaudal. The sulcus spermaticus divides about the
level of the third subcaudal. There is a naked basal pocket present in
all specimens. There seem to be six proximal rows of relatively large
spines while the remainder of the hemipenis is covered with densely
set, smaller spinules. The outer edge of the distal end of the hemipenis
bears a relatively large, smooth apical disc.

Specimens Examined

Liophis andinus.—^ouyiA: Incachaca, AMNH 36012, 36014, 36016,
CM R2777, R2797-98, R2780-81, R2804-08.

1983

Dixon— Systematics of Liophis

133

Liophis reginae macrostoma.— a: “northern,” MHNG
1552.34; Iguazu Falls, FMNH 9252, 9379; Puerto Iguazu, IML 194.
Brazil: no specific locality, MCZ 1991; Agudos, KU 124638; Anhan-
gai, MZUSP 1659; Annapolis, AMNH 62230, 62233; Aruana, MZUSP
2182; Assis, USNM 165586; “Bahia” BMNH 149.80, MCZ 2952;
Buriti, MZUSP 5352; Corumbatai, MZUSP 4007; Crato, MZUSP 5244:
Franca, MZUSP 828; Itaqua, USNM 100752; Pirapora, MZUSP 829;
Posto Diauarum, MZUSP 3691-92, 5512, Sao Felipe, MZUSP 5715;
Sao Luis de Caceres, MZUSP 1370; “Sao Paulo”, MCZ 17809-10,
UMMZ 62817-18; Taubate, USNM 76390; Usina Sinimbu, Manga-
beira, MZUSP 2940-42, 3337; Utiariti, MZUSP 4743; Vespasiano,
MZUSP 2776; Xavantina, BMNFl 1972. 4.20. Paraguay: Asuncion,
BMNH 1930.11.27.202, FMNH 13160; Primavera, BMNH
1956.1.3.34, 1960.1.3.39, 1962.80; Rosario, BMNH 1962.82; 2.7 km
N San Antonio, UMMZ 14321.

Liophis reginae — French Guiana: Maripasoula, MCZ

77512. Guyana: no specific locality, AMNH 3595, 17680; Acarahy
Mountains, KU 69825; New River BMNH 1939.1.1.87-88, Wismar,
UMMZ 77506 (2). Suriname: no specific localtiy, MCZ 16387, RMNH
12510, USNM 5434, 6117; 3 km SE Blakawatra, RMNH 1877; Bro-
kabaka, RMNH 1859; Brokopondo, MCZ 149538, 152599-601; SE
of Carl Francois, RMNH 1841; Lely Mountains airstrip, RMNH 1650;
12 km NE Lely Mountains airstrip, RMNH 1682; Lelydrop, AMNH
104623; Loksihatti, RMNH 771; 13 km W Moengo, RMNH 387;
Republiek, Para Kreek, RMNH 84; R-V Nature Reserve, Foenjue
Island, Coppename River, MCZ 152601; Sam Kreek, AMNH 104621;
Uitkyk, CM 44301; Zanderij, BMNH 1946.4.4.54.

Liophis reginae semilineata. — Bolivia: no specific locality, AMNH
2985, 4440; Abuna, UMMZ 56874; Buena Vista, UMMZ 60721-22,
64153 (2), 67910-16; AMNH 36006, FMNH 35690, 35702; “Beni,
AMNH 101877; Cachuela, AMNH 22523; Cauriaco, AMNH 101875;
Chamblaya, BMNH 1902.5.29.96; Chiquitas, FMNH 195908; Core-
tolla, AMNH 101837; Guayara Merin, USMN 123970; Huachi, AMNH
22473, 22479, 22503; La Perseveranda, AMNH 101833; Las Yungas,
CM 19, 24, 27; Manao, UMMZ 56870-72, 56882; Riberalta, AMNH
22257; Rio Itenez, AMNH 101873; Rio Madre de Dios, UMMZ 56886,
59786; Rio Mamore, AMNH 101835; Rurrinabaque, AMNH 22445,
22522, Sacramento, MZUSP 4151; 10 km E San Antonio, AMNH
101874; 10 km W San Pedro, AMNH 101834; 4 km N Santa Cruz,
AMNH 101836; Santa Rosa, AMNH 101830-32, 101838-40; Trini-
dad, AMNH 101876. Brazil: no specific locality, MNHP 1912-483,
1912-484, 1912-485, 1912-485A, 1912-485B; Abuna, UMMZ 56877;
“Amazonas”, AMNH 14543; north bank of Amazon, three hours
downstream from Leticia, Colombia, LACM 103621; Angra dos Reis,

134

Annals of Carnegie Museum

VOL. 52

MZUSP 2419; “Bahia,” BMNH 62.1.30.57. MCZ 3285, Bahia, AMNH
22281; Barcarena, KU 128102; Barra do Tapirapes, AMNH 93575-
83, MZUSP 3799, 4326, 4336; Belem, KU 127273-74, 124593, 124599,
128103, 140180; 50 km from Belem, MCZ 53200; Boca do Acre,
MZUSP 5754; Cachimbo, MZUSP 3335-36; Calama, MZUSP 5901;
Campo Novo, MZUSP 5924; Caninde, MZUSP 4238-40, 4251, 4276,
4282; Canutama, MZUSP 5762; Foz do Jamari, MZUSP 5905; Hyu-
tanaha, USNM 28945; Igapo-Assu, BMNH 1946.1.4.66; Igarape Be-
lem, AMNH 115019-21; Lago Catu, MZUSP 1826; Manaus, AMNH
64897, CAS 49796; MZUSP 3053; Marajo Island, BMNH
1923.11.9.117-21; Maves, AMNH 89784,91 643-44; Novo Aripuana,
MZUSP 5917; Obidos, MCZ 1204 (2), 2574 (2), UMMZ 56310; Orix-
imina MZUSP 5487; Pacoti, MZUSP 3632; “Para,” BMNH 45.8.25,
MCZ 898, 1171, 2880, 2882, 3292, 5690; Placido de Castro, MZUSP
2556; Porto Velho, MZUSP 3751-52, 4632; Porto Velho, MZUSP
3691-92; Puruzinho, MZUSP 5908; Rio Branco, MZUSP 2555; “Rio
de Janeiro,” MCZ 3944, 6792; Rio Doce, MCZ 17957, MZUSP 830;
Rio Jurua, MZUSP 833; San Carlos, MZUSP 5904; Santa Leopoldina,
MZUSP 832; Santarem, MCZ 1 162 (3), 3682, MZUSP 1250; Tapaua,
MZUSP 5769; Valenga, MZUSP 5812-13. Colombia: “La Selva,”
MCZ 62256; Leticia, AMNH 78986, 95075, KU 124938, LACM
103622, MCZ 48978-80, 57249; Tarapaca, CAS 135343. Ecuador:
no specific locality (or doubtful locality), AMNH 1 52 1 0, 24 1 47, BMNH
89.4.8.1. (2), LACM 2540; Abitagua, FMNH 25812-14, 28025, 28060;
Alpayacu, FMNH 4069; Arajuno, USNM (JAP) 776; Banos, UMMZ
88964; near Banos, UMMZ 92021, 92023; between Banos and Abi-
tagua, UMMZ 92024-27, 92030; Cabeceras del Rio Arajuno, USNM
(GOV) 8304-05, 8310; Cabeceras del Rio Capahuari, USNM (GOV)
7225, 7413; Canelos, USNM (GOV) 7327; Chichirota, USNM(GOV)
7296, 7416; Concepcion, USNM (GOV) 7325; Copataza, USNM (GOV)
7280-82; Dureno, KU 105413; El Topo, BMNH 1912.11.1.36-39;
Lago Agrio, KU 126035; Limoncocha, UIMNH 54665, 61248-51,
61253-56, 63524, 82477-81, 92245, Llunchi, UMMZ 84761; Loreto,
USNM (JAP) 3742, USNM (GOV) 7292, 7324, 7410, Macuma,
UIMNH 62868, USNM (JAP) 8650-52; Mera, KU 98626, 121324-
26, 133533; Montalvo, USNM (GOV) 7287-88, 7408; Mount Tun-
guragna, FMNH 36623; Palmera, AMNH 36042; Paragachi, USNM
(GOV) 7469; 8387; Plan de Milagro, USNM (JAP) 7024; Pucayacu,
USNM (GOV) 7297; Puyo, AMNH 36043-44; Riobamba and Canelos,
AMNH 35926, headwaters of Rio Bobonaza, USNM (JAP) 8613; Rio
Corrientes, USNM (GOV) 7294; Rio Cotopino, USNM (GOV) 7298;
Rio Curaray, FMNH 23486; region of upper Rio Curaray, USNM
(JAP) 3756; USNM (GOV) 7293, 7412; Rio Napo, UMMZ 88966;
Rio Negro, KU 121327-28; Rio Oglan, USNM (GOV) 7419; Rio Pas-

1983

Dixon— System ATics of Liophis

135

taza, UMMZ 88967-68; upper Rio Pastaza, USNM (JAP) 7938; north
bank of Rio Pastaza, between Banos and Abitagua, UMMZ 92022,
92029; Rio Pindo, near Rio Tigre, USNM (GOV) 7409, 7414-15; Rio
Pucuno, USNM (GOV) 6129; Rio Sano, UMMZ 92028; Rio Siquino,
USNM (GOV) 7326, 7417-18; Rio Villano, USNM (GOV) 7411; Rio
Suno, UMMZ 92031; Rio Talin, USNM (GOV) 7295; San Francisco,
UMMZ 88965; San Jose Viejo de Sumaco, USNM (GOV) 7276; Santa
Cecilia, KU 105408, 105414, 109839, 112274, 148352-54, 152517,
158535; “Santiago/Zamora”, UMMZ 82885; Shell Mera, USNM (GOV)
7284-85; Tiputini, USNM (JAP) 8626. Peru: no specific locality, ANSP
3702, 11294-95, 11539, 14282, 14284,41289, 14301, FMNH 40045,
MCZ 16356-57; Achinamisa, AMNH 52937; Ayendama, AMNH
52524, 52729; Balta, LSUMZ 14599-600, 28002-03; ridge between
two Biabos, AMNH 52896; Cashiboya, AMNH 52915, 52866-68;
Centro Union, TCWC 42116-17; 44681, 44579, 44081-85, 40546-
52, 40554; Cerros del Sira, AMNH 104292; Chanchamayo, AMNH
52188, 52691, FMNH 39642, 40639, 152302; Chancharia, USNM
193786; Chiaco-Bagua, MCZ 120277; Chipurana, AMNH 52209,
52218; Contamana, AMNH 52242; Estera Rohyana, AMNH 106268;
Estiron, MZUSP 4378-79, 4384; Fundo Lotta, USNM 196057; Ha-
cienda Pampayacu, MCZ 42404-05, 42418; Huacamayo, AMNH
21160; Igarape Champuia, MZUSP 3344; Indiana, TCWC 44678; Iqui-
tos, AMNH 52053, 52112, 52136, 52161-62, 52206, 52228, 52314,
52331, 52374, 52408, 52423, 52428, 52485, 52497, 52538-39, 52577,
52591, 52593, 52613, 52616, 52634, 52641, 52643, 52659-60, 52663,
52743-44, 52762, 52772, 52876-77, 52957, 52971, 52999, 53002,
53051, 53055, 53150, 53261, 53263, 53334, 53398, 53403, 53500,
54299, 54387, 54847, 54943, 55950, 56041, 56082, 56085, 56107-
08, 56128, TCWC 38221, 39098, 40553, 41751, 44086-87, 44679,
46279, 47147-49, USNM 197272-78; Jenaro Herrera, MHNG 1567.81;
Juliaca, AMNH 2959 (in error); La Divisioria, Sierra Azul, AMNH
107814, FMNH 56163-69; La Mar, Silva, FMNH 39642; Marcapata,
FMNH 59175, 62956; Madre de Dios, FMNH 40039-41, 40234-35;
Mishana, TCWC 42113-14, USNM 197272 (2); Monte Alegre, AMNH
52783; Monte Carmelo, AMNH 55548, 55555, 55622, 55627, 55629,
55634, 55636, 55641, 55645, 55653, 55655-56, 55918; Morona Co-
cha, AMNH 53135; Moropon, TCWC 38 1 29-20, 38222, 42115, 44672-
77, 39101; Moyobamba, BMNH 74.8.4.20, 74.8.4.69; FMNH 5715;
Nazareth, FMNH 5673; Orellana (Reforma), AMNH 52095, 52909,
55670; Oxapampa, FMNH 134479; Pachisa, AMNH 52565; Pacaya,
AMNH 53597; near Palca, FMNH 134470; Pampa Hermosa, Rio
Cushabatay, AMNH 53377, 53403, 53425-26, 53502, 53509-12,
53514-17, 53522, 55450, 55474, 55713, 55720-21, 55739, 55747,
55763, 55755-66, 55770, 55792, 55794, 55798, 55837, 55851, 55854,

136

Annals of Carnegie Museum

VOL. 52

55856, 55858, 55860-61, 55895, 55961, 55982, 55986, 55999, 56005,
56008-09, 56022; Pampa Hermosa, Rio Ucayali, AMNH 52023-26,
52030, 53550, 53576-77; Panya, AMNH 52281, 52347-48; Paraiso,
TCWC 42112; Parinari, AMNH 55680; Parvenio, AMNH 52457; Pe-
tras, ANSP 1 1469, 11671; Pozuzo, FMNH 5579; Puerto Mairo, FMNH
3678; Pucallpa, AMNH 71 157-58; Punga, AMNH 52039, 52041,
52081, 52387, 53482-83, 56045, Quince Mil, FMNH 168384; 40 km
NE Quince Mil, LSUMZ 32553; Quinton, AMNH 21161; headwaters,
Rio Aspusana, AMNH 52096; Rio Itaya, near Iquitos, AMNH 52109,
52111, 52221, 52676, 52802, 53653, 53710, 53176, 53758, 53775,
53811, 53818, 54038, 54073, 54086, 54096, 54136, 54267, 54292,
54303, 54376, 54394-96, 54399, 54401, 54407, 54417, 54419, 54458,
54666, 54669, 54713, 54723, 54726, 54745, 54750, 54778, 54977,
55026, 55054, 55079, 55106-07, 55124, 55158, 55166, 55173, 55194,
56113, 56116; upper Rio Maranon, BMNH 1913.6.4.5-6; upper Rio
Nieva, AMNH 55901; Rio Putumayo, FMNH 37436; Rio Samiria and
Parinari Canyon, AMNH 5271, 57299-302; upper Rio Ucayali, AMNH
71135, 7111-14; Rio Ucayali, UMMZ 51254 (2); Roaboya, AMNH
52235, 52477, 52482, 52799, 52888, 53095, 54433, 55690-91; Sa-
canche, USNM 196056; San Ramon, FMNH 152302; Sarayacu, BMNH
81.5.13.37; Sobral, AMNH 55350, 55345; Suhuayo (Contamana)
AMNH 53007; Suhuaya (Rean Rean), AMNH 53578; Tingo Maria,
USNM (WS) 3113, USNM 193767-85; Tocache Nuevo, USNM
196058; Utuquina, AMNH 52938; Valle de Iscozazin, Chontilla, LACM
76806; Yanamono, TCWC 39099-100, 42118, 44080, 44680; Yari-
nacocha, FMNH 45591, 56123-24, 56145-49.

Liophis reginae zwc//c//. — Guyana: Mabaruma Compound, USNM
164207-10; north slopes of Mount Roraima, BMNH 1971. 1724. Trin-
idad: no specific locality, BMNH 1947.3.3.27, USNM 17757-58; Ar-
ima, AMNH 81462; Arima-Blanchiceusse Rd., AMNH 73136; Brick-
field, FMNH 49958; Mayaro, MCZ 49069; Mount Aripo, BMNH
1940.3.11.85; Rio Grande Forest, Sangre Grande, AMNH 81464,
85953; San Rafael, FMNH 49957; Tamana Caves, Mount Tamana,
MCZ 100654; near Valencia, MCZ 81517; Vega de Dropouche Rd.,
AMNH 85954. Venezuela: Aricagua, BMNH 1905.5.31.55; Arabopo,
UMMZ 85279; Caracas, MHNG 367.56, MBUCV 505; 9 km S Ca-
racas, USNM 196332; Caripito, AMNH 67877-78, 98260-61; Caru-
pano, ANSP 5510; Capibara, USNM (FN) 35446; Cerro Turumiquire,
AMNH 29317, FMNH 17833-36; Cerro Yapacana, RMNH 2276,
2285; Cuchiuano, AMNH 29332; Cumana, UMMZ 56038; El Est-
anque, Borburata, MBUCV 3082; El Junquito, MBUCV 3083, SCN
4318; El Limon, CM 7355; El Vigia, CM 8001; El Yaque, Near Tur-
umiquire, CM 7969; Los Canales, Naiguata, CM 22780; 4 km NW

1983

Dixon— Systematics of Liophis

137

Montalban, USNM (FN) 19641; 2.7 km NE Pena Blanca, KU 167586;
planta electrica de Curpupao, AMNH 59430; Puerto Cabello, UIMNH
93850; Rancho Grande, AMNH 98262, BMNH 1970.237, CAS 138485,
FMNH 204477, KU 167585, MBUCV 95-96, 621, 659, 3076-81,
MCZ 62496, 81518, SCN 15103, UIMNH 22663, 63600, UMMZ
124225-33, 128390; San Antonio de Maturm, MCZ 9979; 5 km S San
Juan de las Galdonas, RMNH 2363; Sorte, Chivacoa, SCN 10551;
Viverios Guayabal, altos de Pipe, MCZ 112411.

Liophis reginae subsp.— Tobago: Pigeon Peak Trace, USNM 228069.

Liophis reginae semilineata X L. r. macrostoma.— AKGEHTmw Rio
Pescado, Oran, IML 115; Salta, IML 600. Bolivia: Chamblaya, BMNH
1902.5.29.96. Brazil: Porto Velho, Rio Tapirapes, MZUSP 3751, 3752,
3797, 4326, 4336, 4362.

Liophis reginae reginae X L. r. zvr^//^/z— Guyana: Berbice, BMNH
53.4.6; Demerara, BMNH 55.8.28, 55.8.28.48; Dunoon, UMMZ 53912,
53968-69; Kartabo, AMNH 18170, 67876; Lama Creek, AMNH
36105; Matali, AMNH 61541; Stabu, FMNH 30959, 30962.

Liophis wzY/zam^/. — Venezuela: Cerro El Avila, UIMNH 63607:
Colonia Tovar, CM 7393, USNM 121206; El Junquito, MCZ 51329;
Rancho Grande, MBUCV 3044, UMMZ 124221, 124224.

Acknowledgments

I wholeheartedly thank those curators who allowed me access to their specimens. The
collections examined, their acronyms and respective curators are; American Museum of
Natural History (AMNH), R. G. Zweifel, C. W. Myers; Academy of Natural Sciences,
Philadelphia (ANSP), Edmond Malnate; British Museum (Natural History) (BMNH),
A. G. C. Grandison, A. F. Stimson; California Academy of Sciences (CAS), A. E. Leviton,
R. C. Drewes; Carnegie Museum of Natural History (CM), C. J. McCoy; Field Museum
of Natural History (FMNH), R. Inger, H. Marx; Instituto Miguel Lillo (IML), R. Laurent;
University of Kansas Museum of Natural History (KU), W. E. Duellman; Los Angeles
County Museum of Natural History (LACM), J. W. Wright; Louisiana State University
Museum of Zoology (LSUMZ), D. A. Rossman; Museo de Biologia, Universidad Cen-
trale, Venezuela (MBUCV), F. Mago L.; Harvard University, Museum of Comparative
Zoology (MCZ), E. E. Williams; Museum d’Histoire Naturelle, Geneve (MNHG), Volker
Mahnert; Universidade de Sao Paulo, Museu de Zoologia (MZUSP), P. E. Vanzolini;
Rijksmuseum van Natuurlijke Historie, Leiden (RMNH), M. S. Hoogmoed; Sociedad
de Ciencias Naturales La Salle, Caracas (SCN), C. de Lima G.; Texas Cooperative Wildlife
Collection (TCWC), M. J. McCoid; University of Illinois Museum of Natural History
(UIMNH). D. F. Holfmeister; University of Michigan Museum of Zoology (UMMZ),
A. G. Kluge, R. A. Nussbaum; National Museum of Natural History (USNM), W. R.
Heyer, G. Zug.

I thank C. J. McCoy, Michael McCoid, Jack Sites, Jr., and Robert Dean for reading
all or parts of the manuscript. I especially thank the administration of Texas A&M
University for allowing me the time and funds to examine some of the type material
in European museums. I especially acknowledge my wife, Mary, who has aided me
unselfishly.

138

Annals of Carnegie Museum

VOL. 52

Literature Cited

Amaral, A. do. 1935. Collecta herpetologica no centro do Brazil. Mem. Inst. Butantan,
9:235-346.

Andersson, L. G. 1899. Catalogue of the Linnean type specimens of Linnaeus’ Reptilia
in the Royal Museum in Stockholm. Bih. t. K. Svenska Vet.-Akad. Handl., 24 (4):
1-29.

Boulenger, G. a. 1905. Descriptions of new snakes in the collection of the British
Museum. Ann. Mag. Nat. Hist., ser. 7, 15:453-456.

Dixon, J. R. 1979. Origin and distribution of the reptiles in lowland tropical rain forests
of South America. Pp. 217-240, in The South American herpetofauna: its origin,
evolution and dispersal (W. E. Duellman, ed.), Monogr. Mus. Nat. Hist., Univ.
Kansas, 7:1-485.

— . 1 983. Systematics of the Latin American snake, Liophis epinephelus (Serpentes,

Colubridae). In Advances in herpetology and evolutionary biology (A. Rhodin, ed.)
Harvard Univ. Press, Cambridge, in press.

Hardy, J. D. 1982. Biogeography of Tobago, West Indies, with special reference to
amphibians and reptiles: A review. Bull. Maryland Herp. Soc., 18:37-142.

Hoge, a. R. 1954. Notas erpetologicas. Una nova subspecie de Leimadophis reginae.
Mem. Inst. Butantan, 24:241-244.

Holm, A. 1957. Specimina Linnaeana i Uppsala bevarade zoologiska sammlingar fran
Linnes tid. Uppsala Univ. Aksskr., 1957 (6): 1-68.

Jan, G. 1 863. Enumerazione sistematica degli ofidi apparteneti al gruppo Coronellidae.
Arch. Zool. Anat. Fisiol., 2 (2):2 13-320.

Jan, G., and F. Sordelli. 1866 (1860-1881). Iconographie generate des ophidiens.
Livr., Milan, premier tome, 15-19.

Lacepede, B. G. E. L. COMTE, DE. 1789. Histoire naturelle des quadrupedes ovipares
et des serpens. Imprimerie du Roi, Paris, tome second, 527 pp.

Linnaeus, C. 1758. Systema Naturae per regna tria naturae secondum classes, ordines,
genera, species, cum characteribus, dilferentiis, synonymis, Locis, Tomus 1. Editio
Decima, reformata. Holimiae: Laurentii Salvii, 824 pp.

Marcuzzi, G. 1950. Ofidios existentes en las colecciones de los museos de Caracas
(Venezuela). Novedades Cientificas (Contr. Ocas. Mus. Hist. Nat. La Salle) (Ser.
Zool.), 3:1-20.

Myers, C. W. 1 974. The systematics of Rhadinaea (Colubridae), a genus of new world
snakes. Bull. Amer. Mus. Nat. Hist., 153:1-162.

Roze, j. a. 1958. A new species of the genus Uwtheca (Serpentes, Colubridae) from
Venezuela. Breviora, 88:1-5.

. 1959. Taxonomic notes on a collection of Venezuelan reptiles in the American

Museum of Natural History. Amer. Mus. Novitates, 1934:1-14.

Shaw, G. 1802. General zoology or systematic natural history. Ill (2). Thomas Da-
vidson, London.

Steindachner, F. 1867. Herpetologische Notizen. Sitzungsber. Akad. Wiss. Berlin, 55
(l):265-273.

Wagler, j. 1824. Serpentum brasiliensium species novae ou histoire naturelle des
especes nouvelles de serpens, recueillies et observees pendant le voyage dans I’in-
terieur de Bresil dans les annees 1817, 1818, 1819, 1820, execute par ordre de Sa
Majesti le Roi de Baviere, publiee par Jean de Spix, . . . Ecrite d’apres les notes du
voyageur par Jean Wagler. Franc. Seraph. Hubschmann, Monachii viii-75 pp.

>

J

i ■■

. V

I

- 1 ^ i

:/

.t t

,ri>, wv., ■ ' v^i

‘ ,7 , =■ ■';■' •

fell

\

•?'

£^iiSM£)

^YPCS'/‘=^

ISSN 0097-4463

ANNALS

o/CARNECiE MUSEUM

CARNEGIE MUSEUM OF NATURAL HISTORY

4400 FORBES AVENUE PITTSBURGH, PENNSYLVANIA 15213

VOLUME 52 17 JUNE 1983 ARTICLE 7

GEOGRAPHIC AND SECONDARY SEXUAL VARIATION IN
THE MORPHOLOGY OF EPTESICUS EUSCUS

Christopher D. Burnett
Abstract

This study describes geographic and secondary sexual variation in the morphology of
the big brown bat, Eptesicus fuscus. The two aspects of morphology examined were wing
size (three characters) and skull size (10 characters). Approximately 5000 specimens were
measured and divided into 97 geographic groups distributed throughout the North Amer-
ican range of the species.

Females were significantly larger than males for 1 2 of the 1 3 characters studies, but
the absolute differences were small (1.3-3. 8%), and both the degree and direction of
sexual dimorphism were geographically variable. The degree of dimorphism was posi-
tively correlated with litter size and moisture and negatively correlated with temperature,
suggesting the importance of wingloading, moisture stress, and thermoregulation as fac-
tors contributing to sexual dimorphism.

Most of the morphological characters differed significantly from a random spatial
distribution, and the geographic patterns of wing and skull variation were randomly
distributed with respect to each other. Wing size exhibited a strong trend toward larger
size at lower latitudes, whereas skull size showed a reverse, but more complex, pattern.
Insular and peninsular subspecies were clearly differentiated, but there was considerable
morphological overlap among continental subspecies. Modification of continental sub-
species boundaries substantially improved discrimination rates. The pattern of mor-
phological differentiation among insular subspecies cannot be explained by a reduction
of gene flow from a Central American dispersal origin.

^ Address: Department of Biology, Boston University, 2 Cummington Street, Boston,
MA 02215.

^ Present address: Illinois Natural History Survey, 607 East Peabody Drive, Champaign,

IL 61820.

Submitted 9 July 1982.

139

140

Annals of Carnegie Museum

VOL. 52

Introduction

Eptesicus fuscus is one of the most widely distributed mammals in
North America, ranging across southern Canada, the continental United
States, most of Mexico and Central America, and the Greater Antilles
(Hall, 1981). It consequently inhabits climatic regions as disparate as
temperate deserts and tropical rain forests. The extraordinary ecological
amplitude of E. fuscus combined with its sedentary movement patterns
(Beer, 1955; Hitchcock 1965; Davis et al., 1968; Barbour and Davis,
1969; Mills et al., 1975) make this species an excellent candidate for
the study of adaptive variation. In addition, the occurrence of several
insular populations allows an assessment of the dilferentiating effect of
geographic barriers to gene flow. Previous studies of geographic vari-
ation in E. fuscus have reported varying degrees of differentiation with-
in geographically limited areas (Allen, 1933; Engels, 1936; Cockrum,
1952; Jones, 1964; Howard, 1967; Long and Severson, 1969), but no
quantitative assessment of variation throughout the North American
range of the species has been made.

The literature also indicates that females of E. fuscus are generally
larger than males (Engels, 1936; Phillips, 1966; Patterson and Davis,
1968), and data on sexual dimorphism in this species have been used
to support both sides of the recent debate over the causes of sexual
size dimorphism in vespertilionid bats (Myers, 1978; Williams and
Findley, 1979).

In this paper, I describe the geographic patterns of morphological
variation throughout the North American range of E. fuscus. I begin
by considering whether sexual dimorphism warrants separate treatment
of males and females and whether patterns of variation in the two sexes
are concordant. Next, the patterns of variation in wing and skull char-
acter complexes are described and compared. Finally, I quantify the
degree of differentiation among subspecies and revise currently rec-
ognized subspecies boundaries to maximize discrimination among the
continental subspecies. The geographic and climatic correlates of mor-
phological variation in E. fuscus are examined in a separate paper
(Burnett, 1983).

Materials and Methods

Data Collection. — I recorded three wing and 10 skull dimensions (Fig. 1) from 4791
adult skin and skull specimens housed in 61 collections in the United States, Canada,
and Mexico (Appendix I). Morphological characters included: 1) measurements that
could be repeated with accuracy, 2) measurements frequently used by mammalian sys-
tematists to allow comparison with other studies, 3) wing and skull characters to allow
comparison of two functional complexes, and 4) the subset of 19 original characters
which best discriminated among six subspecies in a preliminary discriminant function
analysis on 23 widely scattered populations. Measurements were taken with Helios, knife-
blade, dial calipers and recorded to the nearest 0.1 mm. Museum number, collection

1983

Burnett— £'/’r£'S/ct/s Variation

141

Fig. 1 . — Illustration of wing and skull characters of Eptesicus fuscus used in this study:
FA, forearm length; M3, third metacarpal length; M5, fifth metacarpal length; MN,
mandibular toothrow length; HC, height of coronoid process; ON, occipitonasal length;
MX, maxillary toothrow length; PP, postpalatal length; ZB, zygomatic breadth; MS,
mastoid breadth; lO, interorbital breadth; RB, rostral breadth; DB, depth of braincase.
Skull diagrams after Hall (1981). Wing and skull illustrations not to scale.

date, collection locality (to county), sex, and reproductive status of females were recorded
from the specimen labels.

Because my primary purpose was to describe continent-wide trends, I arranged spec-
imens into geographic groups (GG’s) by pooling adjacent collection localities until sta-
tistically meaningful sample sizes were obtained. The distribution and reference numbers
of the resulting 97 geographic groups are shown in Fig. 2. Mean sample sizes (±SE) for
these groups were 20.7 (± 1 . 1) for males and 23.0 (±1.2) for females. Fourteen male and
11 female GG’s were composed of fewer than 10 individuals, but these groups were
retained because of the large geographic gaps that would have otherwise resulted. The
use of small sample sizes for a few localities is further justified by the very low within
population coefficients of variation that bats show for most linear measurements (Long,
1968).

142

Annals of Carnegie Museum

VOL. 52

Fig. 2. — Map of Eptesicus fuscus geographic groups (GG’s) used in this study. All GG’s
were represented by male and female samples except GG 1 1 1 (males only) and GG’s
18, 31, and 1 15 (females only). Total GG’s: male = 94, female = 96, both sexes repre-
sented = 93. Dotted lines indicate subspecies boundaries revised from Hall (1981) on
the basis of discriminant function analyses (see Table 3).

Data — Descriptive statistics on the 13 univariate characters were computed

separately for males and females for each of the 97 geographic groups. The resulting GG
means provided the basis for subsequent analyses. Principal components analyses (PCA)
of the combined male and female means were used to further summarize overall mor-
phological variation. I kept wing and skull variables separate, assuming that they rep-
resent dilferent functional complexes and that pooling them would obscure their differ-
ences. To determine if the directions of greatest variance were uniform among groups
(Neff and Marcus, 1980), principal component analyses were also performed on each of
12 widely scattered GG’s. Student’s /-tests were used to assess the overall degree of

1983

Burnett— £'pr£’5'/ci75' Variation

143

secondary sexual dimorphism using the univariate group means, the principal component
scores, and the ratios FA/ON and M3/M5 (Fenton, 1972). Student’s /-tests were also
conducted for sexual dimorphism in each of 1 3 widely scattered GG’s, and covariance
analysis was used to test for sexual dimorphism in wing size after adjusting males and
females to a common body size.

The first step in the analysis of geographic variation was to test each variable for
departure from spatial randomness. By constructing a matrix of phenetic distances be-
tween all pairs of localities and a comparable matrix of geographic distances. Mantel’s
Z statistic was computed on the null hypothesis of random association between the two
matrices (Sokal, 1979). I constructed phenetic distance matrices in which the elements
were computed as Xij = | — Xj | , where X; and Xj are the values of variable X (mor-

phometric means) at localities i and j respectively. Similarly, I constructed a geographic
distance matrix using map distances (km) between pairs of points. Because some paths
of connection between GG’s cross water barriers which I assumed to be ordinarily
impassable to bats, nine GG’s were eliminated (GG’s 101, 113, 114, 115, 116, 125, 126,
127, 130) along with groups for which data on only one sex were available (GG’s 18,
31, 111, 115) before conducting Mantel’s tests.

To graphically display patterns of variation, I used the “contouring” feature of the
computer program, SYMAP (Dougenik and Sheehan, 1975), to produce isarithmic maps.
Male and female maps were based on 94 and 96 control points (GG’s) respectively, and
the range of data values for each variable was divided into five or six equal intervals.
Impermeable “barriers” were placed around the four island samples and between Baja
California and mainland Mexico to prevent interpolation across bodies of water that I
assumed to be barriers to the movement of E. fuscus.

To assess the adequacy of currently recognized subspecies boundaries, I performed
discriminant function analyses for each sex on a subset of the GG’s using subspecies
boundaries (Hall, 1981) to assign the a priori groups. I then made bivariate plots of the
first principal components of wing size versus skull size for all GG’s and drew envelopes
around the samples of each subspecies. If there was morphological overlap between
geographically distant portions of two subspecies, I made no changes. But if there was
geographic as well as morphological overlap, I reassigned the subspecies of GG’s to
minimize overlap in morphological and geographic space. I then repeated the discrim-
inant function analyses and compared the percent of cases correctly classified obtained
using the modified boundaries with those obtained initially.

All analyses were carried out on an IBM 370/168 computer at the Boston University
Academic Computing Center. Descriptive univariate statistics, /-tests, covariance anal-
yses, principal components analyses, and discriminant function/canonical variates anal-
yses, were performed with SPSS programs (Nie et al., 1975). Mantel’s tests were per-
formed using the GEOVAR series of programs developed by David Mallis (Sokal, 1979).
I modified these programs, written for use on a Univac computer, to be compatible with
the Boston University IBM system. Two-dimensional plots of the principal component
scores were produced with SAS programs (SAS Institute, 1979). Three-dimensional plots
of subspecies centroids on the first three canonical variate axes were generated by a
program I wrote using DISSPLA subroutines (Integrated Software Systems Corporation,
1975). Hard copies of these plots were produced on an FR-80 laser device.

Results

Descriptive Statistics and Sexual Dimorphism

Descriptive statistics based on sample means from the 93 geographic
groups for which data were available for males and females are in Table
1 . Descriptive statistics for each GG and other data not included here

144

Annals of Carnegie Museum

VOL. 52

Table Descriptive statistics and one-tailed \-tests of secondary sexual dimorphism in
Eptesicus fuscus, based on sample means from 93 geographic groups (df = 184). Signif-
icance levels: *,V < 0.05; **, P < 0.01; ***, P < 0.001.

Character

Mean (Minimum-maximum) in mm

cv

9/<3

i-value

Male

Female

Male

Female

Ratio

FA

46.1 (41.9-51.3)

47.2 (42.5-51.3)

3.2

3.0

1.024

5.20***

M3

42.3 (38.1-47.6)

43.4 (39.2-46.9)

3.4

2.9

1.026

5.22***

M5

40.0 (35.8-44.5)

41.0 (36.6-44.2)

3.3

3.1

1.025

5.37***

MN

8.3 (7.4-9. 1)

8.5 (7.5-9.0)

3.0

2.9

1.024

3.50***

HC

5.8 (5. 1-6.5)

6.0 (5. 3-6.5)

3.9

3.7

1.034

6.21***

ON

16.0(14.3-17.5)

16.3 (14.6-17.4)

2.9

2.8

1.019

5.04***

MX

7.9 (7. 1-8.6)

8.0 (7. 1-8.6)

2.9

2.7

1.013

4.16***

PP

7.1 (6.4-7. 6)

7.2 (6.5-7.6)

2.5

2.5

1.014

4.78***

ZB

12.5(11.1-13.5)

12.8(11.3-13.7)

3.1

3.0

1.024

4.85***

MS

9.8 (8.3-10.3)

10.0 (8.6-10.5)

3.1

3.0

1.020

4.34***

lO

4.2 (3. 8-4.4)

4.2(3.8-4.5)

2.7

2.9

1.000

1.23

RB

5.9 (5.0-6. 2)

6.0 (5. 1-6.4)

3.3

3.5

1.017

3.06**

DB

5.2 (5.0-6.0)

5.4 (5.0~6.0)

7.9

9.0

1.038

2.17**

FA/ON

2.88 (2.78-3.07)

2.89 (2.79-3.09)

2.2

2.1

1.003

-1.17

M3/M5

1.06 (1.04-1.10)

1.06 (1.04-1.09)

0.8

0.9

0.999

-0.81

PCIW

-0.39 (-3.18-2.96)

0.32 (-2.62-2.73)

2.66**

PClS

-0.22 (-4.52-1.22)

0.18 (-4.26-1.94)

5.34***

are in Burnett (1982). The first two principal components of skull
variation (PC IS and PC2S) had eigenvalues greater than one (Table 2)
and accounted for more variation than any one of the original variables;
they were retained for subsequent analysis. PC IS accounts for 78% of

Table 2.— Results of principal component analyses of interlocality variation in Eptesicus
fuscus. Wing and skull characters were analyzed separately, and only components with
eigenvalues > 1 were retained.

Character loadings

Wing

PClW

Skull

PClS

PC2S

FA

0.336

MN

0.959

0.090

M3

0.338

HC

0.892

0.296

M5

0.337

ON

0.972

0.113

MX

0.930

0.116

PP

0.908

-0.253

ZB

0.971

0.001

MS

0.906

-0.317

lO

0.676

-0.612

RB

0.947

-0.077

DB

0.591

0.736

Variance explained

97.8%

78.2%

12.1%

Eigenvalue

2.94

7.82

1.21

1983

Burnett— £'Fr£'5'/cj7S Variation

145

SEX DIMORPH. PC I W

0 500 1000

Fig. 3. — Isarithmic (contour) map of the degree of sexual dimorphism in the first principal
component of wing variation in Eptesicus fuscus. Darker shading indicates greater di-
morphism favoring females.

the variation in the 1 0 skull variables and thus provides a good measure
of overall skull size (Humphries et ah, 1981). Only the first principal
component of wing variation (PC 1 W) was extracted by the eigenvalue
> 1 criterion, but it accounts for 98% of the variation in the three wing
characters (Table 2). PCIW thus provides an excellent summary of
variation in wing size. Comparison of the character loadings obtained
from separate PCA’s on 12 widely scattered geographic groups indi-
cated that the directions of greatest variance are similar throughout
the species.

Females are significantly larger than males for all univariate char-
acters except interorbital breadth (lO), which was the same for both
sexes (Table 1). The first principal component scores for wing and skull
characters are also significantly greater in females. The sexual differ-

146

Annals of Carnegie Museum

VOL. 52

SEX DIMORPH. PC I S

Fig. 4.— Isarithmic map of the degree of sexual dimorphism in the first principal com-
ponent of skull variation in Eptesicus fuscus.

ences are small, however, ranging from 1.3 to 3.8% (Table 1). Using
forearm (FA) and occipitonasal (ON) lengths as measures of wing and
body size, respectively, the ratio FA/ON shows that relative wing length
is not significantly greater in females than in males. Likewise, the ratio
M3/M5 shows that relative wing width is statistically equal in males
and females.

Analysis of the group means, however, may not be representative ii
for the entire range of the species. The results of /-tests conducted on ||
13 GG’s selected from diverse localities and subspecies showed con- ||
siderable variation in the characters that are dimorphic. A few cases
of greater male size were also found, indicating that the direction of
dimorphism also varies.

Using the differences between male and female first principal com-

1983

Burnett— £'pr£'5'/Ci75’ Variation

147

Fig. 5. — Isarithmic map of the first principal component of wing variation (PCIW) in
male Eptesicus fuscus.

ponent scores as indices of dimorphism, the geographic patterns of
variation in the degree of wing and skull size dimorphism were mapped
(Figs. 3-4). The pattern of wing dimorphism does not differ significantly
from a random distribution. Nevertheless, inspection of the maps re-
veals two interesting trends: a trend of increasing dimorphism favoring
females toward the east, and a reverse trend of dimorphism favoring
larger males at low latitudes. Based on the significant differences be-
tween males and females and the geographically variable nature of this
dimorphism, I chose to treat the sexes separately in subsequent anal-
yses.

Geographic Patterns of Variation

The results of the Mantel’s tests showed the patterns of variation of
all the wing variables and all but two male (PP, MS) and four female

148

Annals of Carnegie Museum

VOL. 52

Fig. 6. — Isarithmic map of the first principal component of skull variation (PC IS) in
male Eptesicus fuscus.

(MN, PP, ZB, MS) skull variables to be significantly dilferent from a
random distribution. To assess the influence of the spatially random
skull variables on later analyses, I removed the three most random
variables (PP, ZB, and MS), conducted a PCA on the seven remaining [
skull variables, and mapped the resulting first principal component
scores. No major differences were found between the maps based on
the reduced and full sets of skull variables. Because only two variables
were random for both sexes (PP, MS) and because neither of these
dominated the principal component scores, the spatially random vari-
ables were retained.

The geographic patterns of the first principal components of male
wing and skull variation are illustrated in Figs. 5-6. Female patterns
differ in detail, but the general trends are the same. A point distribution

1983

Burnett— £'pr£5'/Ci75' Variation

149

Table 3. — Comparison of discriminant function classification rates for nine subspecies of
Eptesicus fuscus obtained using subspecies boundaries from Hall (1981) (Before) with
those obtained using modified boundaries (After).

Subspecies

Percent of cases correctly classified

Males

Females

Before

After

Change

Before

After

Change

bahamensis

100

100

0

97

97

0

bernardinus

49

69

+20

76

73

-3

dutertreus

—

—

81

81

0

fuscus

82

80

-2

57

65

+8

hispaniolae

—

—

—

100

100

0

miradorensis

88

89

+ 1

85

90

+5

pallidus

29

53

+24

29

33

+4

peninsulae

96

96

0

100

100

0

wetmorei

—

—

—

47

47

0

Overall

67

76

+9

66

70

+4

coefficient of 2.7 was obtained for these maps, indicating that the dis-
tribution of control points (geographic group localities) provides a high-
ly reliable data base for interpolation (Dougenik and Sheehan, 1975).
Nevertheless, the purpose of these maps is to reveal general trends.

Inspection of the map of wing variation (Fig. 5) reveals a strong trend
toward increased size at lower latitudes. The major deviations from
this general trend are small-winged bats in Baja California (GG 101),
the Bahamas (GG 125), and Cuba (GG 126) and large-winged bats in
the northwest, Alberta (GG 89), and northern Wisconsin/Michigan
(GG 51). Large-winged bats also occur in the southwestern United
States.

In contrast to the relatively simple and concordant patterns shown
by the wing characters, the skull characters exhibited a more complex
picture. As represented by the first principal component of skull vari-
ation (PC IS), overall skull size is largest in the northwest and along a
midcontinental band extending from Canada to Central America (Fig.
6). Baja California, the Bahamas, and Cuba again have unusually small
bats. All skull and wing characters for both sexes exhibit a trend of
increasing size eastward across the Greater Antilles.

Subspecies Discrimination

I obtained adequate data to allow discriminant function analyses
(DFA) on six subspecies for males and nine subspecies for females
(Table 3). The following subsets of geographic groups were selected to
represent the subspecies with more than one group: bernardinus, 2, 3,
6, 7, 10, 11, 13, and 15; pallidus, 1, 89, 38, 35, 44, 16, 17, 100, and
\0A\ fuscus, 50, 40, 45, 105, 63, 75, 82, and 85; and miradorensis, 107,

150

Annals of Carnegie Museum

VOL. 52

Fig. 7.— Plot of the first principal component of wing variation (PCIW) versus the first
principal component of skull variation (PC IS) for 96 geographic groups of female Ep-
tesicus fuscus. Circled GG numbers indicate insular and peninsular subspecies (see Fig.
2). Polygons enclose the GG’s for each continental subspecies as defined by Hall (1981).
GG’s 74 and 89 are outliers from E. f fuscus.

108, 106, 109, 110, 112, 113, and 1 14. To maximize sample sizes, the
frequently missing skull character, zygomatic breadth (ZB), was omitted
from the analysis. Preliminary DFA’s did not include ZB as a significant
discriminating variable for either males or females.

Using subspecies boundaries depicted in Hall (1981) as the basis for
the a priori groups, the overall rates of correct classification were 67%
for males and 66% for females (Table 3). The classification rates of
only 29% for E. f pallidus, however, suggested that the boundaries of
this subspecies could be improved.

A bivariate plot of the first principal components of wing and skull
variation for the 96 female groups coded by subspecies showed good
separation among the insular and peninsular subspecies but extensive
morphological overlap among the continental subspecies (Fig. 7). Using
this plot, I made the following changes in subspecies designation: GG’s
1 and 89, pallidus to bernardinus; GG’s 14 and 15, bernardinus to
pallidus\ GG’s 39 and 44, pallidus to fuscus\ GG 104, pallidus to

1983

Burnett— £'pr£-5'/cc/5’ Variation

151

Fig. 8.— Plot of the first principal component of wing variation (PCIW) versus the first
principal component of skull variation (PC 1 S) for females of the four continental sub-
species of Eptesicus fuscus. Polygons enclose the GG’s for each subspecies as modified
by this study (see Fig. 9). Only the morphologically peripheral GG’s for each subspecies
are indicated. GG’s 74 and 89 were outliers.

miradorensis\ GG fuscus to miradorensis. Fig. 2 shows the revised
subspecies boundaries. The corresponding plot of principal component
scores for the continental subspecies shows greatly reduced overlap
between pallidus and fuscus and bernardinus (Fig. 8). In addition, dis-
crimination between fuscus and miradorensis is improved. Classifi-
cation rates were improved in males, increasing by 20% (from 49 to
69%) in bernardinus and 24% (from 29 to 53%) in pallidus (Table 3).
Male miradorensis as well as female fuscus, miradorensis, and pallidus
all improved slightly. The only losses of discrimination were a 3% drop
in female bernardinus and a 2% drop in male fuscus. The complete
classification results produced by the final discriminant analyses are
given for males and females in Tables 4 and 5. For males, four signif-
icant discriminant functions were extracted and 1 1 of the 1 2 eligible
variables were entered in order of decreasing discriminating ability as
follows: MS, DB, FA, lO, RB, HC, MX, ON, MN, PP, and M5 (M3
was not used). Six significant functions were obtained for females and

152

Annals of Carnegie Museum

VOL. 52

FEMALES

SUBSPECIES
A = wetmorel
X = bahamensls
▼ = bernardinus
X = dutertreus
© = fuscus
ffi = hispaniolae
n = miradorensls
• = pal Ildus
+ = peninsulae

canon

Fig. 9.— Plot of the female group centroids of nine subspecies of Eptesicus fuscus on the
first three canonical variate axes.

Table 4. — Classification results of discriminant function analysis on six subspecies of male

Eptesicus fuscus.

Predicted group membership

Group

Cases

baham-

ensis

bernar-

dinus

fuscus

mira-

dorensis

pal-

lidus

penin-

sulae

bahamensis

18

18

0

0

0

0

0

100%

0%

0%

0%

0%

0%

bernardinus

121

0

83

21

5

12

0

0%

69%

17%

4%

10%

0%

fuscus

179

0

23

143

5

8

0

0%

13%

80%

3%

5%

0%

miradorensis

105

0

4

4

93

4

0

0%

4%

4%

89%

4%

0%

pallidus

79

0

14

18

2

42

3

0%

17%

23%

3%

53%

4%

peninsulae

23

1

0

0

0

0

22

4%

0%

0%

0%

0%

96%

Percent of “grouped” cases correctly classified: 76.38%

1983

Burnett— £'P7’£'5'/ct/5' Variation

153

all 12 variables were entered into the analysis in the following order:
MS, DB, RB, FA, lO, HC, MX, ON, PP, MN, M3, and M5.

The degree of differentiation among subspecies is summarized by
plotting the group centroids of each subspecies on the first three ca-
nonical variate axes (Fig. 9). The most distinct populations are the very
small subspecies from Baja California (peninsulae) and the Bahamas
(bahamensis). The Antillean subspecies {dutertreus, hispaniolae, and
wetmorei) are also well differentiated, and it is interesting to note that
their relative positions in morphological space reflect their geographic
locations. The tight clustering of the four continental subspecies (fuscus,
pallidus, bernardinus, and miradorensis) reflects their high degree of
morphological overlap relative to the insular and peninsular subspecies.

Discussion
Sexual Dimorphism

Attempts to explain sexual dimorphism have usually invoked the
concepts of sexual selection (Darwin, 1859; Trivers, 1972) or niche
partitioning (Selander, 1966; Wilson, 1975). It has been argued recently,
however, that for mammals the most common selective pressures fa-
voring large size in females are related to several aspects of reproductive
physiology (Ralls, 1976). As one of the mammalian taxa in which
females are often larger than males, vespertilionid bats provide a suit-
able group for investigating Ralls’ “big mother” hypothesis. Indeed,
two alternative theories have been proposed to explain the advantages
of larger female size in the Vespertilionidae since Ralls’ review paper.

Myers (1 978) proposed that sexual size dimorphism in vespertilionid
bats is related to the extra mass females carry during pregnancy and
lactation. Specifically, he proposed reduction of the proportionate load
of the fetus(es), reduction of the relative cost of producing milk, and
an increase in the quantity of ingested insects females can carry as the
advantages of large size in female bats. Myers made two predictions
based on this wing-loading hypothesis: 1) a greater degree of sexual
dimorphism should be found in females carrying or feeding a greater
mass of fetal tissue or young, and 2) if males and females had the same
body size, female wings would still be larger than male wings.

I tested the first prediction by comparing the degree of dimorphism
in groups of E. fuscus with litter sizes of one versus groups with litter
sizes of two. The embryo data I obtained from specimen labels agree
with previous accounts describing the litter size of E. fuscus as two in
eastern and one in western North America (Cockrum, 1955; Kunz,
1974). I used GG’s 4-38 to represent the west and GG’s 48-83 to
represent the east, omitting GG’s with questionable or variable litter
sizes. Following Myers (1978), I used the difference between average

Table 5. — Classification results of discriminant function analysis on nine subspecies of female Eptesicus fuscus.

154

Annals of Carnegie Museum

VOL. 52

m oo oo oo oo

O O in 'O o
o

o o o o

O CM
(N ^

O O (N oo O O

OO (N^ ^ \o 0^'0 OO

OO OO OO (Ncn

0 0^0
— o

r-- r- o o

o o o o o o

o o o o o o

o o o o

oo fOfN 0^00 oo

o oo oo oo oo

O O 'Ti Tf

O F-

m o^

oo oo oo oo oo oo

00 F'

oo oo oo oo oo oo oo

VO

o —

Percent of “grouped” cases correctly classified: 70.31%

1983

Burnett— £’/’7’£S/ci75' Variation

155

Table 6.— Results of covariance analyses for secondary sexual dimorphism of forearm
length (adjusted by occipitonasal length) in selected geographic groups o/Eptesicus fuscus.
Significance levels: * P < 0.05; **, P < 0.01; ***, P < 0.001; N.S., P > 0.05.

Subspecies (GG)

Adjusted FA (N) in mm

Equality
of slopes

Equality
of adjusted
means

Male

Female

bernardinus (4)

47.0 (29)

48.5 (22)

N.S.

bernardinus (7)

46.0 (23)

46.5 (42)

N.S.

N.S.

bernardinus (8)

46.2 (16)

47.8(18)

N.S.

*

bernardinus (11)

46.8(16)

47.1 (23)

N.S.

bernardinus (13)

46.4 (28)

47.3 (28)

N.S.

N.S.

pallidus (16)

45.2 (27)

46.4(18)

N.S.

*

pallidus (17)

44.7 (19)

45.1 (25)

N.S.

N.S.

pallidus (29)

47.5 (47)

47.5 (27)

N.S.

N.S.

pallidus (30)

47.6 (30)

47.1 (25)

N.S.

N.S.

pallidus (32)

46.9 (25)

47.6 (26)

*

N.S.

pallidus (37)

45.0 (24)

45.5 (36)

N.S.

N.S.

pallidus (102)

45.5 (26)

46.8 (26)

*

pallidus (103)

46.6 (23)

47.2 (23)

N.S.

N.S.

fuscus (41)

45.8 (22)

47.2 (32)

N.S.

fuscus (59)

44.8 (33)

45.5 (24)

N.S.

N.S.

fuscus (61)

45.6 (34)

46.4 (34)

*

♦

fuscus (75)

44.6 (36)

45.3 (37)

N.S.

N.S.

fuscus (82)

44.6 (23)

45.0 (36)

N.S.

N.S.

peninsulae (101)

41.9 (23)

42.3(19)

N.S.

N.S.

miradorensis (108)

48.1 (20)

48.6 (22)

N.S.

N.S.

bahamensis (125)

44.7 (25)

44.8 (30)

N.S.

N.S.

dutertreus (126)

45.5(10)

47.0 (14)

N.S.

♦

Overall means

46.5 (93)

46.8 (39)

N.S.

♦

Overall PCIW, PC IS

-0.3 (93)

-0.1 (93)

N.S.

female and male forearm lengths to measure '‘degree of dimorphism,”
Using a one-tailed Mann- Whitney f/-test (Siegel, 1956), the degree of
dimorphism in E. fuscus proved to be significantly greater in the east
1.25, N = 29) than in the west (X = 1.01,N = 30, 329.0, P =

0.05) in agreement with Myers’ first prediction.

To test Myers’ second prediction, I used covariance analysis (Klein-
baum and Kupper, 1978) to “remove” the effect of body size and to
test for the significance of the remaining difference between the “ad-
justed” mean male and female forearm lengths. Occipitonasal length
(ON), the character with the highest loading on the first principal com-
ponent of skull variation (Table 2), was used as the body size covariate.
The results of these analyses (Table 6) showed the adjusted means for
females to be significantly larger in seven of the 22 GG’s tested. In two
of these cases, however, the assumption of equality of slopes was vi-
olated. Adjusted “wing size” was also significantly greater in females
using univariate (FA adjusted by ON) and multivariate (PCIW ad-

156

Annals of Carnegie Museum

VOL. 52

Table 1 .—Spearman rank-order correlations between degree of sexual dimorphism in
wing and skull size in Eptesicus fuscus and climatic variables. Significance levels: * P <
0.05; **, P < 0.01; all others, P > 0.05.

Climatic variable

N

Wing

(9PC1W-5PC1W)

Skull

(9PC1S-5PC1S)

Mean annual temperature

93

0.10

-0.13

Mean January temperature

93

0.03

-0.20*

Mean July temperature

93

0.07

0.02

Mean daily maximum temperature, January

87

-0.07

-0.26**

Mean daily maximum temperature, July

87

-0.01

-0.07

Mean daily minimum temperature, January

87

0.07

-0.16

Mean daily minimum temperature, July

87

0.20*

0.04

Length of freeze-free period

93

0.03

-0.18*

Actual evapotranspiration

93

0.23**

0.09

Mean annual precipitation

93

0.19*

0.06

Mean relative humidity, January

89

0.24**

0.11

Mean relative humidity, July

89

0.20*

0.00

justed by PC IS) GG mean data. Thus, limited support was found for
Myers’ second prediction.

Williams and Findley (1979) found weaker support for these pre-
dictions in their data than did Myers. Williams and Findley found
adjusted forearm length to be greater in males for the New Mexico
sample they studied. This “reverse” dimorphism was also observed in
my New Mexico sample, GG 30 (probably the same specimens used
by Williams and Findley), but this GG is not representative of the
species as a whole.

Williams and Findley (1979) proposed that the primary factor se-
lecting for larger body size in female vespertilionids is the greater en-
ergetic efficiency of larger body size in maintaining homeothermy dur-
ing gestation. I interpret their hypothesis to predict a greater degree of
sexual dimorphism favoring larger females where cold stress during
the gestation period is greater. Because it is difficult to determine the
operative environmental temperatures (Bakken, 1976) to which bats are
actually exposed even where detailed roost temperature data are avail-
able (Burnett and August, 1981), a rigorous test of Williams and Find-
ley’s hypothesis is not possible. Nevertheless, I correlated the degree
of wing and skull dimorphism with 12 climatic variables because of
the general importance of bioclimatic trends (Table 7).

Skull dimorphism was significantly, negatively correlated with three
of the temperature variables, indicating relatively larger females where
temperatures are colder. Significant positive correlations between wing
dimorphism and the four moisture variables were found and may be
due to sexual differences in roosting sites. In arid regions, any wing-

1983

Burnett— £'pr£’5’/cc/5' Variation

157

loading advantage which females obtain from larger wings may be offset
by excessive evaporation. Males should be less subject to moisture
stress because their roosting sites are probably cooler (Barbour and
Davis, 1969).

The occurrence of geographic variation in the social organization of
several species of bats (Bradbury, 1977) suggests another set of factors
with potential for affecting sexual dimorphism. For example, in low
latitude, nonhibemating populations, it may be possible for male E.
fuscus to obtain sole mating access to females. Perhaps sexual selection
is the cause of dimorphism favoring males at low latitudes.

My results do not resolve the question of which selective forces, if
any, are most important in causing sexual dimorphism, but they do
encourage further investigation of both the wing-loading and ther-
moregulation hypotheses. More importantly, the demonstration that
the degree and even the direction of dimorphism in E. fuscus are
geographically variable constitutes a strong warning that generaliza-
tions derived from geographically limited data are unwarranted.

Geographic Variation and Subspecies Discrimination

To compare the variation of wing and skull character complexes, I
used Mantel’s Z statistic to test the degree of association between the
first principal components of wing and skull variation for each sex.
The phenetic distance matrices of PCIW and PC IS did not depart
significantly from a random association with each other for either sex
(males: t == 1.385, P > 0.10; females: t = 0.760, P > 0.15), emphasiz-
ing that there is no relationship between the geographic patterns of
wing and skull variation in E. fuscus. These results suggest that different
causal factors are responsible for regional trends in these two functional
complexes. The variable chosen to represent “size” in E. fuscus will
thus have an important influence on the outcome of ecological and
evolutionary interpretations regarding body size (Burnett, 1983).

The problem of describing intraspecific variation in E. fuscus is
complicated by the high degree of morphological overlap between pal-
lidus and adjacent subspecies. While the overall rates of correct clas-
sification using the revised subspecies boundaries are 76 and 70% for
males and females, the corresponding rates for pallidus are 53 and 33%
(Table 3). Considering differentiation between pallidus and fuscus first,
the greatest morphological overlap occurs between the pallidus groups
34-38, and the geographically distant fuscus groups, 63, 73, 76, and
77 (Figs. 2 and 8). Conversely, the fuscus groups geographically closest
to pallidus (39-42, 44, and 48) are morphologically distinct from neigh-
boring pallidus. The boundary between pallidus and fuscus is, thus, a
morphologically abrupt one. A more gradual transition appears to exist

158

Annals of Carnegie Museum

VOL. 52

between pallidus and bernardinus, where there is a tendency for the
geographically closer groups of the two species to be morphologically
similar. The transition between pallidus and miradorensis is part of a
long dine along the eastern edge of the range of pallidus which includes
geographic groups 34, 32, 31, 47, 104, and 105 (Figs. 2 and 8). Just to
the west, however, a sharp transition occurs between groups 103 and
104. Another sharp break occurs between the pallidus groups 16 and
17, followed by a long dine of decreasing size along Baja California to
peninsulae (Fig. 7). E.fuscus peninsulae is restricted to Baja California
Sur and is easily distinguished from pallidus of northern Baja California
by its extremely small size. In sum, the geographically central position
of pallidus is reflected in its morphologically central position. The high
rates of misclassification obtained for it, however, can be attributed to
morphological convergence with geographically distant populations as
well as to clinal transitions to adjacent subspecies in several areas.

After pallidus, the continental subspecies with the lowest classifi-
cation rates is bernardinus, with 69% of the males and 73% of the
females being classified correctly (Table 3). This subspecies was most
often confused with fuscus, despite the great geographic distance be-
tween the nearest populations. Few specimens were available from the
Great Basin and Columbia Plateau regions where bernardinus meets
pallidus, but the fact that the geographically closest pair of these sub-
species (GG’s 7 and 1 3) are morphologically similar (Fig. 8) suggests
a gradual transition between these subspecies.

Overlap between the eastern subspecies, fuscus, and pallidus was
described above. Conclusions about the nature of the boundary be-
tween fuscus and miradorensis are prevented by the lack of specimens
from southern Texas, but the morphologically most similar groups are
geographically closest, suggesting a gradual transition. Because few in-
tact specimens were available from the area mapped by Hall (1981) as
the Florida subspecies, osceola, I pooled specimens from this area with
those from the coastal plains of Georgia and South Carolina. Despite
pooling, this geographic group (GG 68) was morphologicaly distinct
from fuscus (Fig. 7). Unfortunately, sample sizes were still too small
to include the osceola samples in the discriminant analyses.

In miradorensis, males and females were correctly classified in at
least 90% of the cases analyzed (Table 3). Variation within this sub-
species forms a dine of increasing wing and skull size from Mexico to
Honduras. The samples from Costa Rica/Panama indicate a slight
reversal of the trend, and the few specimens available from Colombia
suggest that wing size decreases slightly but that skull size continues
to increase into South America.

Of the four insular subspecies from the Caribbean for which I had
sufficient data, all were well differentiated from each other. The sub-

1983

Burnett— £'P7’£'5'/Ci75' Variation

159

species from New Providence and adjacent islands, bahamensis, was
correctly classified in 100% of the male and 97% of the female cases
(Table 3). One female specimen was placed in the Baja California
subspecies, peninsulae (Table 5), which resembles bahamensis in its
very small size. Specimens from Bahamian islands other than New
Providence were assigned to the Cuban subspecies, dutertreus, by
Koopman et al. (1957), but I pooled the specimens from all the Bahama
Islands in the discriminant analyses.

My sample of dutertreus was confused with the neighboring hispan-
iolae in 13% of the cases and with mainland miradorensis in 6% of the
cases (Table 5). Silva Taboada (1979) found no significant differences
in forearm or occipitopremaxillary lengths within samples of dutertreus
from western, central, and eastern Cuba. He did, however, find highly
significant differences between the pooled samples from Cuba and the
small subspecies, petersoni, from nearby Isla de Pinos.

The subspecies from Hispaniola, hispaniolae, was correctly classified
100% of the time, but the Puerto Rican subspecies, wetmorei, was
misclassified as mainland miradorensis in 53% of the cases (Table 5).
In their review of the zoogeography of Antillean bats. Baker and Gen-
oways (1978) argued that E. fuscus colonized the Antilles from the
west, either from the Yucatan peninsula or the Central American main-
land. If this dispersal hypothesis is correct, the observed pattern of
morphological variation strongly suggests that reduced gene flow has
not been the dominant cause of insular differentiation. This is because
morphological distance from the mainland subspecies decreases with
increasing geographic distance eastward across the archipelago (Figs.
7 and 9, Table 5). If reduced gene ffow was the dominant differentiating
force, morphological and geographic distances should be directly, rather
than inversely, related. Considerable difference of opinion exists, how-
ever, regarding the complex historical biogeography and geology of the
Caribbean islands (Pregill, 1981; MacFadden, 1981). Interpretation of
variation among the insular subspecies should, therefore, not rule out
other biogeographic scenarios. For example, the large size of Eptesicus
guadeloupensis (Genoways and Baker, 1975), coupled with its possible
origin from the northern coast of South America (Baker and Genoways,
1978), suggests a route of dispersal that is consistent with a gene ffow
model of differentiation among Caribbean Eptesicus. The possible im-
portance of selective factors in differentiation among insular subspecies
of E. fuscus is discussed in a companion paper (Burnett, 1983).

Acknowledgments

I thank T. H. Kunz and M. A. Bogan for advice throughout the course of this research.
I am also grateful to J. Olson for help with computer cartography; R. Stewart and K.
Mauer for entering data; M. F. Wilcox for illustrations; R. Rosengaus for Spanish trans-

160

Annals of Carnegie Museum

VOL. 52

lation; and M. H. Stack, C. T. Miller, and N. A. Irish for assistance preparing the
manuscript. The curators and staffs of many museums cooperated generously. Partial
financial support was provided by National Science Foundation Doctoral Dissertation
Research Grant DEB-8014703.

Literature Cited

Allen, G. M. 1933. Geographic variation in the big brown bat {Eptesicus fuscus).
Canadian Field-Nat., 47:31-32.

Baker, R. J., and H. H. Genoways. 1978. Zoogeography of Antillean bats. Pp. 53-
97, in Zoogeography in the Caribbean (F. Gill, ed.). Academy of Natural Sciences,
Philadelphia, Special Publ., 13:iii+l-128.

Bakken, G. S. 1976. A heat transfer analysis of animals: unifying concepts and the
application of metabolism chamber data to field ecology. J. Theor. Biol., 60:337-
384.

Barbour, R. W., and W. H. Davis. 1969. Bats of America. Univ. Press Kentucky,
Lexington, 286 pp.

Beer, J. R. 1955. Survival and movements of banded big brown bats. J. Mamm., 36:
242-248.

Bradbury, J. W. 1977. Social organization and communication. Pp. 1-72, in Biology
of bats (W. A. Wimsatt, ed.), Academic Press, New York, 3:1-651.

Burnett, C. D. 1982. Ecogeographic variation in the morphology of the big brown
bat, Eptesicus fuscus. Unpubl. Ph.D. thesis, Boston Univ., Boston, 183 pp.

. 1983. Geographic and climatic correlates of morphological variation in Eptesi-
cus fuscus. J. Mamm., in press.

Burnett, C. D., and P. V. August. 1981. Time and energy budgets for dayroosting
in a maternity colony of Myotis lucifugus. J. Mamm., 62:758-766.

Choate, J. R., and H. H. Genoways. 1975. Collections of Recent mammals in North
America. J. Mamm., 56:452-502.

CocKRUM, E. L. 1952. Mammals of Kansas. Univ. Kansas Publ., Mus. Nat. Hist., 7:
1-303.

. 1955. Reproduction in North American bats. Trans. Kansas Acad. Sci., 58:

487-511.

Darwin, C. 1859. On the origin of species by means of natural selection, or the
preservation of favored races in the struggle for life. John Murray, London, 504 pp.

Davis, W. H., R. W. Barbour, and M. D. Hassell. 1968. Colonial behavior of
Eptesicus fuscus. J. Mamm., 49:44-50.

Dougenik, j. a. and D. E. Sheehan. 1975. SYMAP user’s manual, fifth ed. Harvard
Univ. Grad. School of Design, Cambridge.

Engels, W. L. 1936. Distribution of races of the brown bat {Eptesicus) in western
North America. Amer. Midland Nat., 17:653-660.

Fenton, M. B. 1972. The structure of aerial-feeding bat faunas as indicated by ears
and wing elements. Canadian J. Zool., 50:287-296.

Genoways, H. H., and R. J. Baker. 1975. A new species of Eptesicus from Guadeloupe,
Lesser Antilles (Chiroptera: Vespertilionidae). Occas. Papers Mus., Texas Tech Univ.,
34:1-7.

Hall, E. R. 1981. Mammals of North America. John Wiley and Sons, New York, 1:
1-600.

Hitchcock, H. B. 1965. Twenty-three years of bat banding in Ontario and Quebec.
Canadian Field-Nat., 79:4-14.

Howard, J. A. 1967. Variation in the big brown bat, Eptesicus fuscus, in Kansas.
Unpubl. M.A. thesis, Univ. Kansas, Lawrence, 37 pp.

Humphries, J. M., F. L. Bookstein, B. Chernofe, G. R. Smith, R. L. Elder, and S.

1983

Burnett— £'p7'£’5/cj75- Variation

161

G. Boss. 1981. Multivariate discrimination by shape in relation to size. Syst. Zool.,
30:291-308.

Integrated Software Systems Corporation (ISSCO). 1975. Display integrated soft-
ware system and plotting language manual. Integrated Software Systems Corp., San
Diego, 104 pp.

Jones, J. K., Jr. 1964. Distribution and taxonomy of mammals of Nebraska. Univ.
Kansas Publ., Mus. Nat. Hist., 16:1-356.

Kleinbaum, D. G., and L. L. Kupper. 1978. Applied regression analysis and other
multivariable methods. Duxbury Press, North Scituate, Massachusetts, 556 pp.

Koopman, K. F., M. K. Hecht, and E. Ledecky-Janecek. 1957. Notes on the mammals
of the Bahamas with special reference to the bats. J. Mamm., 38:164.

Kunz, T. H. 1974. Reproduction, growth, and mortality of the vespertilionid bat,
Eptesicus fuscus, in Kansas. J. Mamm., 55:1-13.

Long, C. A. 1968. An analysis of patterns of variation in some representative Mam-
malia. Part I. A review of estimates of variability in selected measurements. Trans.
Kansas Acad. Sci., 71:201-227.

Long, C. A., and R. G. Severson. 1969. Geographic variation in the big brown bat
in the north-central United States. J. Mamm., 50:621-624.

MacFadden, B. j. 1981. Comments on Pregill’s appraisal of historical biogeography
of Caribbean vertebrates: vicariance, dispersal, or both? Syst. Zool., 30:370-372.

Mills, R. S., G. W. Barrett, and M. P. Farrell. 1975. Population dynamics of the
big brown bat {Eptesicus fuscus) in southwestern Ohio. J. Mamm., 56:591-604.

Myers, P. 1978. Sexual dimorphism in size of vespertilionid bats. Amer. Nat., 112:
701-711.

Neff, N. A., and L. F. Marcus. 1980. A survey of multivariate methods for system-
atics. Privately published. New York, 243 pp.

Nie, N. H., C. H. Hull, J. G. Jenkins, K. Steinbrenner, and D. H. Bent. 1975. SPSS:
statistical package for the social sciences. McGraw-Hill, New York, 675 pp.

Patterson, P., and W. H. Davis. 1 968. Variation in size among adult Eptesicus fuscus.
Trans. Kentucky Acad. Sci., 29:1-4.

Phillips, G. L. 1966. Ecology of the big brown bat (Chiroptera: Vespertilionidae) in
northeastern Kansas. Amer. Midland Nat., 75:168-198.

Pregill, G. K. 1981. An appraisal of the vicariance hypothesis in Caribbean bio-
geography and its application to West Indian terrestrial vertebrates. Syst. Zool., 30:
147-155.

Ralls, K. 1976. Mammals in which females are larger than males. Quart. Rev. Biol.,
51:245-276.

SAS Institute Incorporated. 1979. SAS user’s guide. 1979 edition. SAS Institute
Inc., Raleigh, North Carolina, 494 pp.

Selander, R. K. 1966. Sexual dimorphism and differential niche utilization in birds.
Condor, 68:113-151.

Siegel, S. 1956. Nonparametric statistics for the behavioral sciences. McGraw-Hill,
New York, 312 pp.

Silva Taboada, G. 1979. Los murcielagos de Cuba. Editorial Academia, La Habana,
Cuba, 423 pp.

SoKAL, R. R. 1979. Testing statistical significance of geographic variation patterns.
Syst. Zool., 28:227-232.

Trivers, R. L. 1972. Parental investment and sexual selection. Pp. 136-179, in Sexual
selection and the descent of man (1871-1971), (B. S. Campbell, ed.), Aldine-Ath-
erton, Chicago, 378 pp.

Wilson, D. S. 1975. The adequacy of body size as a niche difference. Amer. Nat., 109:
769-784.

Williams, D. F., and J. S. Findley. 1979. Sexual size dimorphism in vespertilionid
bats. Amer. Midland Nat., 102:113-126.

162

Annals of Carnegie Museum

VOL. 52

Appendix I

Summary of Specimens Examined

The number of specimens examined are enclosed in parentheses following the collec-
tion numbers and their acronyms (Choate and Genoways, 1975).

1. UAMZ(21)

17. ROM (107)

33. NAU(22)

58. SDSNH(41)
78. UCONN(80)

84. AS (57)

96. UIMNH(235)
108. MHP(60)

144. MSU(42)

162. MOU(50)

178. UNHP(6)

191. NYSM(12)
211. OU(13)

245. TCWC(147)
274. UPS (183)

283. UWZM(21)

2. UBC(37)

22. ENCB(36)
37. UA(54)

64. MVZ(372)
79. DMNH(40)

85. TTRS(IO)

98. JMM(81)
115. KU(565)

148. UMMZ(126)
170. VMKSC(38)
184. MSB (228)
195. NCSU(6)
219. CM (144)
248. TTU(62)
276. WWC(12)
285. UWSP(26)

3. BCPM (28)
25. UAMC(30)
38. ASUZC(88)
68. UCLA (56)
81. USNM(361)

86. FSM(27)

99. ISU(33)

120. UKEN(41)
155. MMNH(34)
171. UNSM(41)
186. AMNH(194)
200. CMNH(65)
221. ANSP(47)
252. MALB(45)
279. WVMSC(18)
This study (158)

15.

NMC (66)

32.

MNA (50)

41.

CAS (100)

75.

UCM (35)

82.

ABS(l)

91.

FMNH (72)

107.

UNI (8)

122.

LSUMZ (21)

157.

FACM (40)

177.

DCM (15)

188.

CU (24)

209.

OSU (90)

238.

MWU (26)

259.

UU (37)

281.

MPM (6)

Total

1 = 4,791

7 73 I

ISSN 0097-4463

ANNALS

0/ CARNEGIE MUSEUM

CARNEGIE MUSEUM OF NATURAL HISTORY

4400 FORBES AVENUE * PITTSBURGH, PENNSYLVANIA 15213

VOLUME 52 16 SEPTEMBER 1983 ARTICLE 8

RESOURCE EXPLOITATION IN AMAZONIA:
ETHNOECOLOGICAL EXAMPLES FROM
FOUR POPULATIONS

Eugene Parker *

Darrell Posey^

Research Associate, Section of Man

John Frechione
Curatorial Assistant, Section of Man

Luiz Francelino da Silva^

Abstract

Delineation of Amazonian ecological typologies has become more sophisticated as
research in the region progresses. Scientific understanding of Amazon ecosystems, how-
ever, is still regarded as far from sufficient. Existing ecological typologies of the Ama-
zonian biome are discussed and evaluated and examples of ethnoecological knowledge
systems from four Amazon populations are presented. It is suggested that folk knowledge
systems represent an important source of information for scientists concerned with
typology elaboration and for planners interested in the formation of development strat-
egies. It is concluded that there exists an urgent need for further ethnoecological inves-
tigations of indigenous populations in Amazonia.

Introduction

The development of the Amazon region of South America, partic-
ularly within the Brazilian portion of Amazonia, has progressed at a

' Address: Department of Geography, University of Maryland Baltimore County, Bal-
timore, MD 21228.

^ Address: Laboratorio de Etnobiologia, Universidade Federal do Maranhao, Sao Luis,
Maranhao, Brazil, 65.000.

^ Address: Faculdade de Educa^ao, Fundagao Universidade de Amazonas, Manaus,
Amazonas, Brazil, 69.000.

Submitted 22 March 1983.

163

164

Annals of Carnegie Museum

VOL. 52

rapid rate since 1970. Interest in, and enthusiasm for, the development
of the region has been prompted more by extra-regional economic
imperatives than by specific information regarding the region’s resource
base. Development policies and programs for the Amazon region have
been designed and implemented despite a general consensus among
scientists that our understanding of Amazon ecosystems (both macro
and micro) is far from sufficient. This paper will briefly outline existing
ecological typologies employed by researchers in their investigations
of the Amazon Basin, suggest some limitations associated with these
typologies, and present specific examples (drawn from four ethno-
ecological investigations) of indigenous (Amerindian and caboclo) con-
cepts of ecological zones and ecosystemic relationships within zones.

Ethnoecology has been defined as the study of indigenous perceptions
of natural divisions in the biological world and of plant-animal-human
relationships within these divisions (Posey, 1981). These cognitively
defined ecological categories do not exist in isolation; thus, ethno-
ecology must also deal with the perceptions of interrelatedness between,
as well as within, “natural” divisions. The study of natural phenomena
by means of ethnoecological analysis has proven to be a rich source of
ecological data that is directly relevant to modern development strat-
egies and planning (Brokensha et ah, 1980; Posey, 1983). The principal
purpose of this paper is to demonstrate that indigenous and folk per-
ceptions of the environment constitute a grossly underutilized source
of information about Amazonia, and that modern science and tradi-
tional knowledge must be linked if we are to ensure rational (and
sustainable) development and productivity in the Amazon Basin.

The Setting

The Amazon Basin (Amazon region, Amazonia) encompasses an
area roughly 7,000,000 km“ in size within which is found the largest
tract of tropical rainforest in the world. It has been estimated that 20
percent of the Amazon’s 557 million ha of rainforest has already been
destroyed (UNESCO, 1978:22). In the Brazilian Amazon alone, ap-
proximately 100,000 ha are being cleared annually (Barbira-Scazzoc-
chio, 1980:iii). The extent and rapidity of such deforestation has ini-
tiated a cycle of soil compaction and erosion, the destruction of nutrient
cycles, and flooding (Sioli, 1973:323; Gentry and Lopez-Parodi, 1980;
Moran, 1981:4). The resulting water pollution, with its associated
changes in pH, suspended and dissolved loads, and turbidity, reduces
riverine productivity and threatens aquatic life (Schubart, 1977; Love-
joy and Schubart, 1981:21). Species endangerment and extinctions re-
sulting in large part from habitat destruction very likely will become
a regular feature of the ecological reality of Amazonia. Indeed, Gottlieb

1983

Parker et al.— Amazonian Ethnoecology

165

(1981:23) has estimated that 90 percent of the natural inventory of
organisms will become extinct before even basic genetic and taxonomic
descriptions can be made.

Human populations in Amazonia have suffered signihcantly from
efforts to develop and “conquer” the region. Entire Amerindian groups
have been eliminated (Indigena, 1974; Davis, 1977); in just the last 75
years, at least 87 Amerindian groups have become extinct in the Bra-
zilian portion of Amazonia alone (Ribeiro, 1970:238). Caboc/os, the
rural peasantry of Amazonia, have experienced tremendous change
resulting from the direct and indirect effects of development including
the disruption and breakdown of village life, destruction of subsistence
systems, and pernicious poverty (Parker, 1981). Where development
has penetrated directly into traditional caboclo areas, the local residents
have frequently been labelled squatters, expelled (often forcibly) from
the lands they have worked decades if not generations, and forced into
dependency relationships in favelas and cities (Velho, 1972; CNBB-
CEP, 1976; Pinto, 1977; Wood and Schmink, 1979).

With the extinction of each indigenous population and the destruc-
tion of each caboclo community, humanity loses a priceless accumu-
lation of practical knowledge about life in, and adaptation to, tropical
ecosystems. The signihcant human and ecological price paid for Am-
azonian development has yielded few rewards; most development proj-
ects remain viable primarily because they continue to receive govern-
ment subsidies (Goodland, 1975:40; Mahar, 1979). The Brazilian
government has encouraged, and often supported through a variety of
incentives, immense development projects (for example, the Jari sil-
viculture effort in northeastern Amazonia), but almost without excep-
tion they have failed to generate anticipated prohts. For example, roughly
85 percent of the recently established cattle ranches in the Paragominas
region of the Brazilian state of Para are already unproductive due to
pasture degradation (Hecht, 1981:96).

One underlying cause for the difficulties that have beset “modern”
development in Amazonia has been a tendency among observers to
generalize about the ecology of the region. Amazonia has too often
been viewed as a region of vast homogeneous landscapes. Ironically,
these perceptions of Amazonian homogeneity have yielded diametri-
cally opposed schools of thought regarding the development of the
region. On the one hand, Amazonia is seen as an enormous “counterfeit
paradise” in which development threatens the fragile balance of the
Amazon rainforest ecosystem (for example. Meggers, 1971; Goodland
and Irwin, 1975). The opposing school views Amazonia as a vast region
of untold resource wealth, its sheer size and homogeneity (in terms of
the rainforest ecosystem) the best argument against those who fear that
signihcant, and/or irreversible, damage will result from development

166

Annals of Carnegie Museum

VOL. 52

activities (for example, Rebelo, 1973; Pandolfo, 1978; Schmithusen,
1978). While the propensity to view the Amazon as a homogeneous
landscape is understandable given the assumed isomorphism between
the region and the humid rainforest, it nevertheless overlooks the fact
that it is an incredibly diverse region with numerous ecological and
micro-ecological zones.

Western science has become increasingly sophisticated in its under-
standing of the complexity of Amazonian ecology, but indigenous and
folk knowledge systems are still generally ignored as sources of valuable
ecological information. Recent publications by Brokensha et al. (1980)
and Klee (1980) provide evidence, from a worldwide perspective, of
the importance and applicability of indigenous knowledge systems in
modern development planning and program assessment. Indigenous
ecological knowledge can offer important lessons for sustained pro-
ductivity in Amazonia. Unfortunately, sustainability has not been a
feature of development programs in contemporary Amazonia.

Amazonian Ecological Typologies

Broad scale scientific interest in the Amazon region only began during
the last century; consequently, ecological typologies are of ra ther recent
vintage. Early classificatory descriptions of Amazonia were, to say the
least, very general. Wissler (1917) simply classified all of the Amazon
Basin surrounded by the Brazilian and Guiana shield complexes under
the general label “manioc area” (cited in Moran, 1981:7). Steward’s
seven volume Handbook of South American Indians (1946-59), a
standard reference, grouped the myriad Amerindian groups together
under the category “tropical forest cultures” and described much of
the Amazon region and its inhabitants as “marginal,” with little or no
evidence of “advanced” agriculture or civilization.

The degree of generalization employed in ecological descriptions of
Amazonia remained, for the most part, unchanged until the middle of
the present century. Beginning around 1950, and continuing until quite
recently, researchers have predicated their discussions of Amazonian
ecology upon the distinction between terra flrme and vdrzea environ-
ments (De Carmargo, 1949; Wagley, 1953; Denevan, 1966; Lathrap,
1970; Meggers, 1971; Sternberg, 1975). The terra firme, lowlands of
Amazonia not subject to inundation, constitute about 98 percent of
the Amazon Basin. The vdrzea (floodplain) is estimated to cover no
more than 2 percent of Amazonia. The importance of the terra firine-
vdrzea dichotomy is directly related to the commonly held belief that
the soils of the terra firme are extremely impoverished and thus in-
capable of sustaining intensive agriculture without massive capital in-
puts. The vdrzea, on the other hand, is perceived as an important
ecological zone and potential development area due to its annual re-

1983

Parker et al.— Amazonian Ethnoecology

167

juvenation that results from flooding and subsequent deposition of
nutrient-rich sediments. Though the distinction between these two
landscapes is a logical one, it nevertheless is of only marginal utility
in the elaboration of ecological typologies because it treats vast areas
as undifferentiated landscapes.

Denevan (1976), unsatisfied with the terra firme-vdrzea division,
was one of the first researchers to attempt, systematically, a detailed
typology of the Amazonian biome. He proposed a classification system
in which six categories were articulated: (1) humid forest; (2) seasonal
forest; (3) montane forest; (4) dry savanna; (5) wet savanna; and (6)
floodplain. Despite this improvement in typological detail, it should
be noted that the humid (tropical) rainforest was still treated as a
homogeneous entity. Fires (1974) has identified two types of terra firme
humid iovQsX—mata pesada (mature upland forest) and ?nata de cipo
(liana forest). Prance (1979) has recognized three categories of humid
terra firme forest in his classification system of Amazon vegetation: (1)
high forest with large biomass; (2) liana forest; and (3) low forest with
reduced biomass.

The treatment accorded the vdrzea landscape is illustrative of the
fledgling nature of ecological typologies developed for the Amazon
region. Early descriptions divided the floodplain into two categories—
"^vdrzea"" was employed for areas that only experienced seasonal in-
undation, whereas "dgapo"' was used to denote permanently inundated
areas (De Carmargo, 1949; Lima, 1956). This scheme prevailed until
Sioli (1967, 191 5a) suggested a more detailed description of floodplain
structure in which the principal elements were: (1) main river; (2)
natural levees (restingas); (3) channel islands; (4) parands (side chan-
nels); (5) floodable grasslands; (6) vdrzea lakes; (7) sand bars (point,
convex, barranco); and (8) igapo (back swamps). Although Sioli spe-
cifically developed this structural scheme for the lower Amazon flood-
plain, it has often been employed by others as a standard description
for all floodplain areas in Amazonia. Parker (1981) has demonstrated
the need for further articulation of floodplain typology due to significant
variations in fluvial dynamics found within and among floodplain zones
in Amazonia. Caboclos, the principal inhabitants of floodplain envi-
ronments in contemporary Amazonia, recognize four different types
of vdrzea based upon the source (cause) of flooding: (1) vdrzea de chuva
(rain); (2) vdrzea de rio (river); (3) vdrzea de mare (tidal); (4) vdrzea de
mar {sQ2i) (Sombroek, 1966).

Geochemical investigations of fluvial environments within Ama-
zonia have identified three distinct types of rivers— white-water, black-
water, and clear-water (see Gibbs, 1967, and Sioli, 1968, for further
discussion). This differentiation of river types has contributed to con-
siderable terminological confusion in ecological typologies of the vdr-

168

Annals of Carnegie Museum

VOL. 52

zea, particularly with regard to the use of igapd. While some observers
(Irmler, 1977; Sioli, 1975Z?; Smith, 1981) have limited the use of iga-
pd'' to black-water areas (regardless of the duration of flooding), others
have applied it to both clear- and black-water areas (Pires, 1974; Sioli,
1975a). Sioli (1975<2:212) has also used igapd to refer to the perma-
nently flooded terra firme side of vdrzea lakes. The issue is yet further
complicated by the fact that igapd is still employed by some observers
to denote permanently flooded areas (Sternberg, 1975:18).

Prance (1979), in an eflbrt to reduce the confusion, has suggested a
biogeographical classification of inundated areas that includes: (1) sea-
sonal vdrzea— forest flooded by regular annual cycles of white- water
rivers; (2) seasonal forest flooded by regular annual cycles of

black- and clear- water rivers; (3) mangrove — forest flooded twice daily
by salt-water tides; (4) tidal vdrzea— forest flooded twice daily by fresh-
water backed up by tides; (5) floodplain forest— forest irregularly flood-
ed by rainfall; (6) permanent white-water swamp forest; and (7) per-
manent igapd, or black-water forest. Parker (1983) has noted that this
classification scheme may foster additional confusion because it only
deals with inundated areas that are covered by forest — an important
feature of the middle and lower Amazon floodplain, as well as portions
of the estuarine floodplain, is that they contain extensive areas covered
by grasses (Lima, 1956; Fittkau et al., 1975; Sioli, 1975a, 1975d; Stern-
berg, 1975; Smith, 1981).

It is evident from the foregoing discussion that progress is being
made in our understanding of, and appreciation for, the ecological
heterogeneity of Amazonia. Knowledge of ecosystem structures and
dynamics continues to be enhanced as research results are made avail-
able (Denevan, 1982; Hecht, 1982). However, it is our belief that the
extent of ecological/biological heterogeneity in the Amazon region can-
not be fully appreciated without consideration and scientific investi-
gation of folk knowledge systems of Amazonian peoples. Recent in-
vestigations by such researchers as Reichel-Dolmatolf (1976, 1978),
Smole (1976), Carneiro (1978), Posey (1979, 1981), Vickers (1979),
Beckerman (1980), Moran (1981:41-57), Chernela (1982), and De-
nevan et al., (1982), support this belief by the demonstration of the
sophistication, diversity, and richness of folk knowledge about Ama-
zonian flora, fauna, and ecological relationships. The goal of this paper
is to provide further evidence in support of the argument that folk
knowledge systems are of crucial importance to the development of
the Amazon Basin, and that they must be studied and incorporated
into scientific and development enterprises in the region.

Following are brief examples of folk ecological knowledge systems
derived from investigations of four Amazon communities. Two indig-
enous societies, the Kayapo of Brazil and the Yekuana of Venezuela,

1983

Parker et al.— Amazonian Ethnoecology

169

both situated on the terra firme, and two caboclo communities, Coari
in western Amazonia and Limoeiro do Ajuru of eastern Amazonia,
both located within floodplain areas {vdrzea), will be discussed (Fig.
1). Thus, the two major adaptive systems found in Amazonia— Amer-
indian and caboclo— Mid the two principal geographic landscapes of
the region— the terra firme and the vdrzea— mq represented.

The Kayapo:

Long-term Resource Management and
Semi-domestication

The Kayapo Indians occupy a two million ha reserve in the Brazilian
states of Para and Mato Grosso do Norte (Fig. 1). The approximately
2500 Kayapo live in nine widely dispersed villages. The discussion
presented here is based upon fieldwork conducted in Gorotire, the
largest of these villages (Posey, 1979).

Within the territory of the Kayapo, three major environmental di-
visions are found— kapdt (grassland), krai (mountains), and bd (forest).
Major ecological zones are recognized within each division (Table 1),
and within zones a number of discrete ecological categories are further
distinguished. Ideally, villages are situated near as many different eco-
logical zones as possible to maximize resource diversity. Near the vil-
lage of Gorotire, for example, the Kayapo recognize eight major eco-
logical zones and two types of transitional zones (Fig. 2).

Specific plants and animals are associated with each ecological zone.
The Kayapo possess an intimate knowledge of animal behavior and
know which animals are associated with particular plants. In turn, plant
types are linked with soil types. Thus, each ecological zone is an in-
tegrated system of plant— animal — soil relationships. Table 2 sum-
marizes selected systemic relationships for one of the major ecological
zones of the Gorotire area— the bd-rdrdrd (natural forest with inter-
mittent openings).

The Kayapo, despite having relatively stationary villages, are none-
theless semi-nomadic, spending 4-5 months away from the village each
year. They have evolved an intriguing system of “forest fields” to ensure
the availability of foodstuffs during extended periods away from the
village. A variety of wild plants, most of which grow naturally in the
bd~rdrdrd, are collected as the Kayapo travel through the forest, and
a number of these are replanted near established campsites (usually
within defecation areas). At least 54 species of wild plants, the majority
of which are tuberous, are used by the Kayapo in this replanting scheme.
Posey (1982) has proposed that this replanting in the proximity of
campsites is a form of plant semi-domestication. The Kayapo are thus
able to obtain adequate food with minimal effort while away from the
village by harvesting these “forest fields.”

170

Annals of Carnegie Museum

VOL. 52

Locator map showing the four study areas.

1983

Parker et al.— Amazonian Ethnoecology

171

Table Major ecological zones recognized by the Kayapo.

Kayapo designation

Major zone

Subzone

Characteristics

Kapot

Kapot-kem

Kapot-kamepti
Kapot-kam-boiprek
Pyka-ti |6] krai

grassland, savanna

short grass lands

savanna with tree stands

high grass lands

savanna with intermittent trees

Krai

mountains

Ba

ba-kamrek

ba-epti

ba-kati

ba-rarara

forest

gallery forest
dense jungle
high forest

forest with intermittent openings

The Kayapo folk taxonomic system reveals a preference for “tran-
sitional” ecological zones that grade between two or more semantic
(named) zones. The village of Gorotire is situated amidst one of these
transitional zones which enables the inhabitants to exploit the maxi-
mum species diversity that such zones support (Fig. 2). The propensity
for a “graded continuum” rather than mutually exclusive ecological
categories is also reflected in the Kayapo perception of forest and gar-
den. Abandoned Kayapo gardens are thought to replicate the natural
open forest (bd-rdrdrd) common to the general area. Kayapo slash/
burn gardens are never really abandoned as is often assumed of most
indigenous agriculturists. Though production from domesticates plunges
after two or three years, a variety of plants and fruits continue to be
collected— yams and taro for 5-6 years, bananas {Musa spp.) for up to
15 years, urucu {Bixa orellana) for 20 years or more, and cupa {Cissus
gongylodes) for at least 30 years. Old gardens also provide a variety of
foods that attract wildlife— for example, porco do mato {Tayassu pec-
an), coati {Nasua nasua), deer, paca {Cuniculus paca), agouti {Dasy-
procta spp.), and a wide variety of bird species. The Kayapo, aware of
the attractiveness of old gardens for wildlife, have purposefully dis-
persed their gardens so as to effectively manage game populations over
a large area. Perhaps of equal importance, plants in the natural refores-
tation sequence are extremely important sources of medicinals.

The Kayapo also manipulate animals other than mammals. For ex-
ample, nine species of stingless bees (Meliponidae) have been identified
which may be considered semi-domesticates (Table 3). Table 4 lists
principal species of Apidae utilized by the Kayapo along with their
uses and distinctive traits— these are only a small portion of the total
number of bee types recognized by the Kayapo.

172

Annals of Carnegie Museum

VOL. 52

LEGEND:

Ba-rarara

Forest w/intermittent openings

Ba-kati
High forest

Ba-kamrek
Gallery forest

Ba-§pti
Dense jungle

Kap6t-kgm
Short grass land

Kap6t-kamepti
Savanna w/tree stands

Pyka-ti i6i krai

Savanna w/intermittent trees
Krai

Mountains
Transition zones

Mixed transition zones

Fig. 2. — Ecological zones surrounding the village of Gorotire as perceived by the Kayapo,

Table 2. — Selected soil-plant-animal relationships in the bd-rdrdrd.

1983

Parker et al.— Amazonian Ethnoecology

173

eg ^

e

e 8

^ <u
<U 3

<1-!

N

^ .s

a. N

M M %
o o ^

-S 33 -S

X) X)

o

-c!

p 6

I S'

a'

X o

o ^

^ T3
OJ -S

tC X Ci (U P W)

Q

U. Q u
u u m
<

C3 a

o
c

Q .2?

^ s <u

O CX; (D

o

s; p

a (u 2

> <u m ^ >

>

•S _r X ^ ^ -2 o o
•Tl'O'O p Px P P
3J <u

73 13 o "p

■S -S .s

‘G ^'G

(U CJ o

S Sx G

1/5 (/5

(U (U
> >
p p
-2

T3 T3

Q Uh
U X
<

<0

O

O

S

. ^
a

lA Q

l.l

05

K X
G too
Q 5c
■;^ ^
*3 53 o,

t§|

•S 2^,

•2 S «

^ p ;s:

^ U Oh U U <0

s,

1%

» s

o o

X X >> X X X

X X u C

^ /<i«s

(U!U(U(U <,fflOQ

Ph O E

tortoise,
red paca.
red agouti,
deer,
tapir.

174

Annals of Carnegie Museum

VOL. 52

Table 3>.— Semi-domesticated (manipulated) species of Apidae utilized by the Kayapo.

Kayapo designation

Scientific designation

Ngai-pere-y*

Apis millifera

Ngai-ny-tyk-ti*' ^

Melipona seminigra cf pernigra

Ngai-kumrenx'’ ^

Melipona rufiventris flavolineata

Ngai-re*

Melipona compressipes cf. fasciculata or afinis

mykrwat'

Frieseomelitta sp.

udjy*’ ^

Trigona arnlthea

kukraire'- ^

Trigona dallatorreana

mehnora-kamrek^

Trigona cilipes pellucida

mehnora-tyk^

Scaura longula

' Species whose nests are systematically raided in subsequent seasons.
^ Species whose nests are taken to the village.

^ Species that are encouraged to build nests in dry posts in the houses.

Natural resource concentrations (“resource islands”) are also known
and periodically exploited by the Kayapo. There is an intentional as-
sociation between resource islands, “forest fields,” and campsites, as
is revealed in Fig. 3. Visits to “forest fields” and resource islands, other
than those prompted by hunting trips, are most often determined by
ceremonial cycles.

This brief but representative survey of Kayapo ethnoecological
knowledge suggests there is a great deal that both the scientific and
developmental communities can learn from the Amerindian about
ecological zonation and ecosystemic relationships within and between
zones. Furthermore, resource concentrations can be identified, includ-
ing those intentionally manipulated to increase productivity of semi-
domesticated plants and wild animal species. The perception of agri-
culture as part of a natural continuum with the tropical forest is perhaps
the most valuable lesson from the Kayapo because of its implications
for long-term resource management and utilization of gardens and
forest resources.

The Yekuana:

Garden Site Selection and
Configuration

The Yekuana Indian community of Aseneenya is located in southern
Venezuela (Fig. 1) within the broad expanse of terra flrme that blankets
the Guiana Shield (a pre-Cambrian crystalline rock complex covering
much of northern Amazonia). While the location of the community is
not within the catchment area of the Amazon River, the area is included
in general discussions of Amazonian ecosystems, particularly the rain-

1983

Parker et al.~- Amazonian Ethnoecology

175

Fig. 3. — Resource islands and campsites associated with “forest fields.

Table ‘X.— Principal species of Apidae utilized by the Kayapo.

176

Annals of Carnegie Museum

VOL. 52

<

•n

S §

t: -S
a S

.5 o

^ 5b

fS «5

« e

C

£ o

™ §
^5 g S V-

S

S i S) S

^ - X) - -

t:

03 'P

a §

<U JS
V
Xi

™ g I -2. .'S, 30

= § f ji - 3

■s 2 & ^ §

<u c

1/3 -rt

3 ^

C3 £3
<U (U

C3 TJ

^ ^ ^ ^

1) 3

.3 M

_ C3

^ £

-.3 '.3 c3

e s

•S -

V

>

■ 3 1/3
<U

3

I I

■(-• C3

C ^ _C

CJ <U

2 -1

'O

a

d

d

d

Cl.

^ s:

a

i/3

(/3

a

Q

"S

d

§

^ o
§ g

■i i

s:

.§)

S'

•§>

s

.§)

X

s:

.§)

S

C3,

a

s

.§3

(/3

a

.§)

a

s:

,|)

O

o

o

O

-t3 £ 'C3

00 C3 00
C .id c

00 g .JL

g f ‘S)

.d^ £ c

Table 4. — Continued.

1983

Parker et al.— -Amazonian Ethnoecology

177

a (/)

c

Ofi
O >

■ S «

C ^

o

3

X

cs

i-f

« 3
M i-

O ,o
6^

w

aj ^
M «

O S

d N
ea S N

“ il

e §

t u

U

d d

s'-i

^ fi

d o

O

U U

''aj 'm

d

ttJ '!>

E S

3

CsJ

&Q

« «

00 00

O

e

<

>

little

avera

much

cd

Ix

li <u
> >

«s cd

little

little

little

little

Season

a!i year

dry

dry

all year

ail year
all year

dry

dry

2? &
xj -d

S

+ +

+ +

+

+ +

U

+ +

+ +

+

+ +

P

+ +

+ +

+

+ +

-S

o

3

•2

1

c

•H

c

bj

■ S^

'E! K

?

s

2 ■>

■a

u

d

d

d d

5: s

dj *3

O S
bo ^

S

c

■|

S -C
1 1,

"E

e

'o

S

o

S

S

a

K

S)

S

s s;

^ §3

§ 5

a <3

S d

§3 §3

5 5

s

o

dj

•§

<3

§3

5

« a
K c

§3 ^

S 5

d

ki

>? S

Y Kl

'g

E a

4>

<V

13

a

dj

a c

1) OJ

C .C

2 d

6 > a

(U D d 5

S ^
^ ^ «

O U C 3

5 ^ ^

a u

S « a

^ 2 >'

ils

£ o.^

> g 'w

!i " " o ^ II II 11

ii ^ m m nj ^ u "

178

Annals of Carnegie Museum

VOL. 52

forest ecosystem. Often considered an area of relatively poor soils and
scarce resources, this terra firme zone is successfully exploited by the
Yekuana who base their subsistence upon shifting cultivation, hunting,
fishing, and gathering.

The Yekuana practice a system of shifting cultivation that includes
site selection, forest clearing, burning, cropping, and fallowing (see
Fuchs, 1 964, and Frechione, 1 98 1 :52-92, for more detailed discussions
of each of these stages). Like other Yekuana subsistence pursuits, the
successful implementation of their cultivation system requires in-depth
knowledge concerning microenvironments, natural resources, and eco-
logical relationships. In order to illustrate the nature of this type of
knowledge, the criteria employed by the Yekuana in garden site selec-
tion, and the subsequent locations and combinations of crops grown
within gardens, will be described.

The primary factor influencing the agricultural system of the Yekuana
is the sharply defined distribution of annual precipitation. In the Ase-
neenya area, 90 percent of the annual rainfall occurs between April
and October. Garden sites must be selected, cleared, and burned before
the onset of the rains or they cannot be cropped that year. A complex
classification of edaphic and vegetative characteristics is employed by
the Yekuana in selecting garden sites. Knowledge of microenviron-
ments with suitable edaphic and vegetative characteristics is obtained
by the Yekuana on frequent hunting and gathering expeditions into
the surrounding tropical forest. Discounted as garden sites are the
following areas: (1) savanna areas— savanna grasslands do not provide
sufficient vegetation for burning, and the soil is difficult to penetrate
with digging sticks; (2) moriche palm forest— areas with moriche palm
{Mauritia flexuosa) stands are considered “too wet” due to poor drain-
age; (3) other wet, poorly drained areas in the forest; (4) forest areas
located on slopes of more than 45°; and (5) sacred areas used as ce-
meteries. There are no other sanctions, spiritual or sacred, regarding
garden locations; however, if a person is killed during garden-making
activities, the garden site cannot be used.

In areas not discounted by the above factors, soil type remains the
primary determinant. The Yekuana word for soil, and earth in general,
is mono. The Yekuana distinguish soil types based upon color and
texture. Six soil colors are recognized: (1) judumato nono—h\2ic\i soil;
(2) seweichato nono— red soil; (3) seweichechato nono— brown soil; (4)
tajededato nono— wbiXe soil; (5) seenato nono— ochre soil; (6) takaje-
yato nono— yeWow soil. Four soil textures are noted by the Yekuana.
They are: (1) sadadajato nono— snndy soil; (2) adinajato tzotzo— clayey
soil; (3) tajucuidaijiato nono— rocky soil; (4) sichuwetojano nono—
wormy soil.

Another indicator used by the Yekuana in locating favorable garden

1983

Parker et al. — Amazonian Ethnoecology

179

sites is the composition of the standing vegetation. Indicators of good
soil are:

1) Sinatajano—Sin area where a good deal of vines and lianas are
present.

2) Suduchijano—a. forest area where a considerable amount of bam-
boo is encountered.

3) Nawdyujano—an area where the wild plantain is found.

4) Yadiwomajano— an area where the Yadiwoma tree grows. This
is a large tree which is used for making doors, tables, and benches.
It grows in both good and poor soils. Hence, it is only a partial
indicator of soil quality and it is necessary to consider other
variables in areas where Yadiwoma grows.

5) Kariyejano— an area where the Kariye tree grows. It generally
indicates soil suitable for musaceous crops.

The existence of a small, running stream is considered favorable
when choosing a garden site, especially if it serves to keep the area
well-drained. A stream also provides the “coolness” considered nec-
essary for some crops.

For agricultural purposes, a number of general microzones based
upon a combination of soil color, texture, and microclimatic factors
are identified by the Yekuana. Each cultivated plant is considered to
grow best when propagated in a particular soil type (Table 5). Overall,
a black, slightly sandy soil is preferred for gardens. This soil type
corresponds to the perceived requirements of manioc, which are always
given paramount attention by the Yekuana (Frechione, 1981:92-98).
A single garden site, however, frequently encompasses a number of soil
types, each of which is planted with those cultigens best suited to the
soil and microzonal characteristics. This results in a patch intercropped
planting pattern in which different crops occupy patches, or zones,
within the gardens (Fig. 4). Mixed intercropping, the practice of inter-
planting two or more crops simultaneously within some patches/zones,
also occurs if the crops share similar preferred microzonal character-
istics (Table 6)— and if other considerations do not discount the in-
tercropping of the plants. However, when mixed intercropping does
exist, it is often for specific crop combinations and usually occurs only
for a short period of time during the garden’s productive lifespan. For
example, if manioc is to be cultivated in a garden, it is always planted
first in the appropriate section (or sections). Corn, which has similar
preferred microenvironmental characteristics as manioc, is then inter-
planted among the manioc mounds. Squash and tania, which also share
similar preferred locational characteristics, are often sparsely inter-
planted within the manioc section. Corn is harvested after about three
months and is not replanted. Squash is harvested in four months, and

180

Annals of Carnegie Museum

VOL. 52

Table 5.— Major cultigens of the Yekuana and preferred sod/microclimate characteristics

for their cultivation.

Cultigen'

Soil color

Soil texture

Micro-

climate

Black

Red

Brown

Yellow

Sandy

Clayey

Wormy Humid^

manioc

1

2

3

X

plantain

1

3

2

X

X X

corn^

1

X

pineapple

1

2

X

X

sugar cane

1

X

X

X X

papaya

anywhere

tania

1

X

yam

1

X

sweet pototo

1

X

squash

1

X

topiro

anywhere

watermellon

1

X

X

chile pepper*

bottle gourd

anywhere

tobacco'*

jute

1

X

X

cotton

1

X

Numbers indicate order of preference for soil color.

‘ See Table 7 for scientific identifications.

2 That is, an area bordering a running stream.

^ Corn should also be planted beside logs left after burning. These logs provide nutrients
as they decay and protect the corn seedlings.

These cultigens should be planted in areas with a lot of ash from burning, or at the site
of a partially or completely burned stump.

tania about eight months after planting. Consequently, within approx-
imately nine months, the manioc section becomes in effect a mono-
cultural patch within the garden, and remains so for the remaining
productive period of the garden (usually from two to four years). Other
cultigen patches exhibit a similar pattern of temporal utilization, with
a dominant, long-term crop and a number of sparsely interplanted
secondary short-term crops.

The Yekuana cultivate at least 76 different named varieties of food
plants, representing some 15 different genera (Table 7). This great
variety is an attempt to maintain the genetic diversity of staple food
crops to insure adaptability to environmental variability (Hames, 1 980:
20). Furthermore, varietal performance in each soil type and drainage
condition is known by the Yekuana. The microenvironmental char-
acteristics of gardens, therefore, determine which of the varieties will
be planted. Consequently, the composition of gardens varies greatly in

1983

Parker et al.— Amazonian Ethnoecology

181

crop complexes (Table 6), and the gardens are not simply tangles of
many different types of randomly interplanted crops.

Yekuana gardens generally range in size from 0.3 to 4.0 ha. Although
monozones are found within gardens (Fig. 4: Sections A and C), entire
monocropped gardens do not usually occur (Frechione, 1983). Yekuana
gardens have been shown to attain manioc yields in excess of 30 tons
per ha (Frechione, 1981:101). This high productivity is undoubtedly
due to successful microzonal planting and the maintenance of high
varietal diversity in cultigens. Thus monozoning, when geared to mi-
croenvironmental variations in soils and climates, can be productive
in a mixed garden system. Monocropping on a larger scale, however,
is considered to be environmentally degrading with resulting high risks
of crop failures. Nonetheless, the Yekuana example offers evidence
that, when properly managed, monozoning can be adapted to ecological
heterogeneity in the Amazon Basin. This brief explication of Yekuana
criteria for site selection, and the microenvironmental conditions best
suited for cultigen propagation, provides one example of the in-depth
knowledge of an indigenous population concerning the exploitation of
their natural environment. It also suggests the degree of complexity of
exploitative strategies that are necessary for the utilization of Ama-
zonian ecosystems in a sustained-yield manner.

The Caboclos of Coari:

Resource Units and
Vertical Levels

The community of Coari is located within a vdrzea (floodplain) zone
at the juncture of Lake Coari and the Amazon (Solimoes) River (Fig.
1). Terra firme zones, however, are found interspersed throughout the
vdrzea in this particular area. Life for the caboclos of Coari centers
around the seasonal changes of water levels in the lakes, rivers, and
streams near the community. Water levels are highest during June when
maximum flooding usually occurs, and are lowest in October. The rainy
season extends from January to April and the dry season from June to
November— December and May are transition months. The average
annual difference between high and low water levels in the Coari area
is 10 meters. It should be noted, however, that unusually high flood-
waters occur on average every 7 to 10 years. Although these high,
damaging floods are infrequent and erratic, floral and faunal distri-
butions, as well as human activities, are limited by their occurrence.

The caboclos of Coari gain their subsistence from a mix of shifting
cultivation, fishing, hunting, and the collection of forest products. Most
agricultural activities are carried out in terra firme areas scattered
throughout the local vdrzea. Caboclos often plant part of their crop on
the vdrzea to take advantage of the nutrient-rich alluvium of the flood-

182

Annals of Carnegie Museum

VOL. 52

LEGEND:

A

Manioc

with

corn; yams

1

Bottle gourd; tobacco

B

Manioc

with

corn; yams; squash

J

Papaya; corn

C

Plantain

with

corn

K

Chile pepper

D

Plantain

with

corn; squash

L

Tobacco

E

Plantain

with

corn; bottle gourd

M

Tania

F

Plantain

with

corn; tobacco

N

Papaya

G

Sugar cane

0

Tupiro

H

Fa'ada

P

Weeds

Fig. 4. — Planting patterns in a typical Yekuana garden.

1983

Parker et al.— Amazonian Ethnoecology

183

plain. However, because erratic flooding can result in the loss of crops
planted in the vdrzea, caboclos will not risk putting their entire agri-
cultural effort into floodplain areas. Thus, the perception of a natural
hazard, that is, flooding, has resulted in a conscious decision by caboclos
to plant most of their crops on the less productive terra firme lands.

Hunting, fishing, and collecting take place in the forests, rivers, lakes,
and streams near Coari. The area near the community encompasses
numerous microecozones that the caboclos recognize and selectively
exploit. Indeed, caboclos possess an intimate knowledge of local eco-
logical variables and recognize at least 40 different types of “resource
units” (lugares de fartura) where important resources are concentrat-
ed—either naturally or due to human manipulation. Fig. 5 shows the
Coari area with some of the locations of these 40 resource units. Table
8 provides a glossary of the numbered units appearing in Fig. 5, and
Table 9 describes some of the general characteristics of these resource
units.

Within many of these resource units, caboclos also differentiate re-
source variability based upon verticality. Terrestrial/arboreal and
aquatic “vertical levels” are recognized by local residents (Fig. 6). Dif-
ferent resources are found at different vertical levels, and resource
distribution between levels also changes with season. Such knowledge
enables the caboclo to locate desired resources and to know what tech-
niques will be required to exploit them. For example, fish and turtles
found in aquatic level A-1 (see Fig. 6) can be exploited with arco e
flecha (bow and arrow), the arpdo (harpoon), the rapiche (net with
handle), or the zagaia (three-pronged spear), while those found in level
A-5 are exploited with the camuri (buoy with hook and line), the jatecd
(turtle harpoon), or the linha comprida (hook and line). Also indicated
for terrestrial/arboreal vertical levels in Fig. 6 are “intermediate zones.”
These zones are interstitial movement corridors for birds and mammals
and are the principal levels for sighting and hunting game.

Game is pursued in relation to the vertical level(s) occupied. Birds
like the jurutis, inambus {Tinamus braziliensis), and bacurau, as well
as anteaters (tamanduas) are found in level T-3 of the castanhal (Table
9) because their preferred foods (insects such as formiga and suava)
are most abundant at that level. Certain hirds— graunas {Gnoremopsar
chopi) and murures—ave found in levels T-3 and T-4 because of the
prevalence of taracuas (another type of insect) in these levels. Other
hivds—japos, japiins {Cacicus icteronotous), araras {Amazon festiva
and Ara macao among others), periquitos (various species) and pa-
pagaios—ntsX during the rainy season at level T-4 in the castanhal.

A complete description of caboclo knowledge concerning even one
resource unit is beyond the scope of this paper (for more detail see
Posey et al., 1983). However, in order to indicate the extent of eth-

184

Annals of Carnegie Museum

VOL. 52

Table 6.-— Patches/zones, crops, and areas in six Yekuana gardens.

Patch/zone

Dominant

Intercropped plants

% of area

Area in
square
meters

Site 1 (.88 ha, see Fig. 4)

A

manioc

corn; yams

55.4

4874

B

manioc

corn; squash; yams

3.8

334

C

plantain

corn

20.7

1819

D

plantain

corn; squash

0.6

51

E

plantain

corn; bottle gourd

0.6

51

F

plantain

corn; tobacco

5.1

445

G

sugar cane

7.4

655

H

fa’ada

1.4

124

I

bottle gourd; tobacco

1.0

91

J

papaya; corn

0.6

51

K

chile pepper

0.6

51

L

tobacco

0.5

48

M

tania

0.4

31

N

papaya

0.2

16

O

tupiro

0.1

9

P

weeds

1.5

128

Totals

100.0

8778

Site 2 (.50 ha)

A

manioc

corn

50.0

2500

B

pineapple

corn

21.0

1050

C

pjlantain

corn

20.0

1000

D

tania

corn

5.0

250

E

sugar cane

4.0

200

Totals

100.0

5000

Site 3 (.90 ha)

A

manioc

tania; cashew

66.0

5940

B

plantain

18.5

1665

C

plantain

sweet potato

0.34

30.6

D

pineapple

13.6

1224

E

chile pepper

0.9

81

F

structure

0.66

59.5

Totals

100.0

9000

Site 4 (3.35 ha)

A

manioc

corn; squash; tania; yams;

85.0

28,475

papaya; chile pepper

B

plantain

corn

11.0

3685

C

pineapple

3.0

1005

D

sugar cane

1.0

335

Totals

100.0

33,500

5’zYF5(1.3ha)

A

pineapple

42.0

1105

B

pineapple

sweet potato

3.3

429

C

pineapple

bottle gourd

1.2

156

1983

Parker et al. — Amazonian Ethnoecology

185

Table 6. — Continued.

Patch/zone

Dominant

Intercropped plants

% of area

Area in
square
meters

D

plantain

23.0

2990

E

plantain

sweet potato

8.2

1066

F

manioc

8.5

1105

G

manioc

sweet potato

0.92

119.6

H

sweet potato

7.5

975

I

sugar cane

2.0

260

J

tania

1.7

221

K

tania

sweet potato

0.6

78

L

adawata akido

0.04

5.2

M

cashew

0.04

5.2

N

weeds

1.0

130

Totals

100.0

13,000

Site 6 (.5 ha)

A

plantain

tania; papaya

87.0

4350

B

plantain

squash

6.0

300

C

sugar cane

2.0

100

D

weeds

sugar cane; papaya

5.0

250

Totals

100.0

5000

noecological knowledge possessed by the caboclos of Coari, a descrip-
tion of some of the interrelationships between resource units 6 {cha-
vascal) and 17 (castanhal) will be presented.

A close relationship exists between the chavascal and castanhal and
it is one of critical importance to the Coari caboclos. During the dry
season, numerous animals come to the chavascal from the castanhal
in search of water that can be found in the small pools and streams
left by receding floodwaters. Small birds and mammals are also found
in the chavascal, lured by the dry season vegetation that flourishes
along the damp beaches (praias) that rim the chavascal. Likewise,
wading birds are found in abundance, attracted by the multitude of
small fish trapped in the shallow waters. Birds such as massaricos
{Hoploxysterus cayanus) and gaivotas, and turtles (for example, ca-
bequdo) lay their eggs in the white sand beaches near the chavascal.
Their eggs attract lizards, such as the jacuraru {Tupinambis nigro-
punctatus), as well as man. Carnivores such as the jaguars {onqas) and
eagles (gavido) move into the chavascal during the dry season attracted
by the many small mammals and birds. It is worth noting that caboclos
recognize that birds such as the cujubim {Pipile cujubi) and small mam-
mals like the paca (Cuniculus paca) act as seed disseminators— these

Manihot esculent a CvdLXWz manioc yuca kiyedi 30 vegetative 10.4 2-4

Musa paradisiaca plantain plMano fadudu 8 vegetative 10.4 3-4

186

Annals of Carnegie Museum

tL, Uh Ph PU, PU Uh

m (N m m (N m

^ ON

TO T3 TO

(U <U (U
OJ (U (U
(/3 (/5 cyi

<U <U (U

PpPLhPhPpPhPh^;^^^^^

—I I

(N

CT3 CT3 ^ (L) (U O 00 OJ

(U't3t3t3t3t3t3

<U <U (U <u

(U ID (U <U <U

00 00 00 00^'^^'^'^'^^'^

O <U <U (U (U ^

(u <u (u o aj

c/2 c/5 (/5 (/) (/)

ON

m H On c6

<u

>

CD

>

'O o

(L» 00

<u <u

c/2 >

O <U OJ
00 00 00
(U QJ (U
> > >

OJ
00

>

'ccfoorxicNrNimc^rNi^rNimrN)— 1^— iNO—i^rxi — i

O

TO g

O

c^ ^

a c

3 ^

c/2 CS

O ... ^

£ '3 g iS iS -£ >
C c3 ^ -.0! ^ iS

I ii'.s

SC3 2 g a El C3

(C 'C ^ ^

ns -2 c« o
o O- o O

Sc^

<C X)

C3 —

a TO

I

1=^

c: 03

o ^

O ... 'a

? ==
> CD

b cas

3 ^

E

E .Q O

■> TO 2
> O g
O 00 -O

2

> 03
03 03

CD 03

E

Eh E

a ^

3 E .2 C E

5= 2? c 5 5

‘E E g3 E C3

E. c/2 ■*— >

E3 ^

c/5 ^

E E TO

O o' E

O c/2

E. o

E

I E

1 E O O
X) .E, O Ei

<U

G & js

%

■Si ts K
^ S Q

a ^ s

^ s

§ i: Q «

“S

Q ^ s;

^ ^ Q
au

2?

^ o o o

’Ll ^
cn Co _
O O

5 Q Q
Cl.

I’S 5

a 'u !.
^ s

I

s

s;

^ 03

G

CO

K

I -

« *3

I «

•2 s:

CO ^-2

C ^ 8
^ ^

VOL. 52

g g

1 3

E o

O >H

B <u

E -E
TO

<U Q
c/2 cE

3 E

X ^
aj 'E
X ^

2 3

TO

(U CD

n

^ to c/5 W

03
DD 1/2

E

C« _
B E

° -a

_ E

s

O (D
>

ON 03

E

217 ^ -2

E

E I

03 O

00 c

03

X -H
o

c« E

E O
E

2 ^
X ^

. (N
TO m

43 ^

43 E
2 £
o g

^ >

TE

B c«
.2^
■X 43

£|
03 ^
00 03
•X 03

S)i

+. E
O CP

^ O

2 <1^
S ’B

V. Oh.
03 2

£•'=5

E 43
E _>

E g
X 2

03 E
03

03 X

00

2

TO

E

d-^

O ^
C (U

o

E P

X ^

u X

^ E g
o .B .o
ax '£

a ^ %

00 .. 7
^ E I

X E 03

2 15

E "S "e

E . 03

■OnX

E

s ^ f

^11
X P .■

03 ^*5 43

)-H 03 ^

'3x2

^ <D CD

03 X fE

p X E

£ X E

X 3 £

'B-S2 I

•g-§x
3 a O

a c/2 jO

x£ I

H ^

* * 43

■» * U2

* *

•K * X

1983

Parker et al. — Amazonian Ethnoecology

187

Fig. 5. — Location of major resource units in the Lake Coari area, and low-high water areas.

188

Annals of Carnegie Museum

VOL. 52

Table S. — Glossary of resource units appearing on Fig. 5, as defined by a cabocio of the
Lake Coari region (numbers correspond to those presented on figure).

1 . praia branca— dry season, white sandy beaches of Lake Coari where birds and turtles
lay their eggs.

2. praia suja~dry season, wet or muddy beaches where a great number of birds feed.

3. praia verde— dry season beaches, covered in short vegetation, where birds feed on
weeds and insects.

4. restinga— natural river levees, usually covered in forest, and not inundated during
the dry season.

5. charco— swamp area found within the varzea ecological zone.

6. chavascal— transition area between rivers draining into Lake Coari and the lake
itself, characterized by low vegetation which is mostly inundated during the rainy
season and which during the dry season forms a labyrinth of dead-end river-like
branches, ressacas, po^os, and pocinhos surrounded by large areas of muddy land.

7. igapo — forest area which is flooded during the height of the rainy season.

8. laguinho— small lake connected to the river by a narrow stream during the rainy
season, and only accessible by land during the dry season.

9. lago grande— large lake, such as Lake Coari or Lake Mamia.

10. lago— lake connected to the river by a passage navigable by canoe or small boat.

1 1. costa— margin (bank) of the Amazon River.

12. enseada— gulf-like section of a large lake, usually characterized by calm waters.

1 3. encontro das aguas— point where the Lake Coari water system flows into the Amazon
River.

14. pogo grande— deep section in a sharp turn of a smaller river, characteristic of the
sinuosity of these rivers.

15. ressaca— lake-like formation connected to a small river.

16. igarape— a black water stream flowing from deep in the forest to a river.

17. castanhal— the terra firme forest where the catanheiras (Bertholetia exceltia) are
located.

18. aguas fundas brancas— deep white waters of the Amazon, where the piraibas {Bra-
chyplatystoma spp) are caught.

19. aguas fundas pretas— deep water areas of Lake Coari, associated with scarcity of
resource except when in close proximity to the banks of the lake.

20. barreiras— high vertical banks of the Amazon River, characterized by swift currents
and an abundance of clay varieties.

21. embaubal— section of the varzea where embauba trees {Cecropia spp) are predom-
inant.

22. buritizal — a concentration of buriti palms (Mauritia flexuosa).

23. jauarizal— a concentration ofjauari trees {Astrocarium jauari) usually at critical zones
during periods of medium water level.

24. agaizal — a coneentration of agai trees {Euterpe edulis) in the terra firme.

25. bacabal— a concentration of bacaba trees {Oenocarpus distichus) along igarapes.

26. aratizal— a concentration of arati (?) bushes at the critical zone during periods of
medium water level.

27. capoeira alta— an abandoned garden site more than ten years old.

28. capoeira baixa— an abandoned garden site approximately five years old.

29. baixio— shallow section of the Amazon River opposite the channel side, character-
ized by the predominance of oeirana trees {Salix martiana).

30. tabocal— a concentration of green-and-yellow bamboo {Guadua spp) in the high
varzea.

1983

Parker et al.— Amazonian Ethnoecology

189

Table — Continued.

31. canaranal— a floating meadow dominated by canarana {Panicum spectabile), com-
monly used as cattle fodder.

32. muriruzal— a floating meadow dominated by muriru (numerous species, see Smith
1981:14), providing food for both fish and turtles.

33. canal seco— navigable channel in the lake during low water.

34. boca de cima (lago)— point where a lake narrows into a stream before entering a
river.

35. boca de baixo (rio)— point where the water from a lake flows into a river.

36. matupazal— a floating meadow dominated by matupa (?).

37. rogado novo— a recently planted slash-and-burn garden site.

38. rogado velho — a slash-and-burn garden site which is still being systematically uti-
lized.

39. igarapezinho— a clear water rivulet which provides drinking water, also called a
fonte.

40. pocinho— a small lake which does not dry up during periods of low water, usually
located near ressacas and chavascals, and where fish are easy to catch by hand.

animals feed on the seeds of chavascal vegetation and through their
droppings or patterns of storage spread the seeds to other areas.

During the rainy season, the chavascal is flooded and its resources
become available to large fish. Caboclos exploit these fish using deep-
water circular throwing nets, bows and arrows, and poles with line and
hook. They also collect arati {Anona densicoma) fruit in their canoes.
The primary focus of caboc/o activity during the rainy season is, how-
ever, the castanhal. Here, the caboclos collect castanhais (Brazil nuts)
for subsistence and for market. They also hunt the mammals and birds
that have been forced, by the high waters in the chavascal, to the
castanhal.

In addition to seasonal variations of floral and faunal resources in
the chavascal and castanhal, there are daily movements, particularly
of birds, into and out of these units. Massaricos {Hoploxysterus cay-
anus), rnarrecos, and mergulhdes {Ibycter formosus) move between
these two resource units at regular intervals during the day. In the early
morning, mergulhdes and garqas {Herodias egretta) fish in the lakes
and ponds near the chavascal, then move to the open areas to sun and
dry themselves in the noon heat, and finally return to their roosts in
the castanhal in the evening.

Though but a quick sketch of the knowledge that Coari caboclos
possess regarding resources and ecological variations, it nonetheless
indicates that caboclo ethnoecological knowledge represents an im-
portant resource to those concerned with developing the Amazon re-
gion.

Table 9. — Some general characteristics of the resource units encountered in the Lake Coari region (numbers correspond to those presented

on Fig. 5).

190

Annals of Carnegie Museum

VOL. 52

0^ 0^

Pi p^:

^ ^ ^ "2.

H' H' H'

HH H' H H H X cti H' H H

H" H" H" hh ^

^ pQ*' m pa pa"' oa" u; u; u; oa'' aa"

pa pa" pa" pa" pa" u;

P3

oa" < < < < < pa" pa" pa" < <

<" <" <" <" <" H H pa"

^ s

> o

m ^ ro f<^

tn in -ciT

rn m rn rn fT <! rn

nf (T nf (N fN fN fN nf <N (N rn fT r4

fN <N (N — rN .-T — I —I — I — c — . -H o4'

HHHHHHH<<<<H <<HHH

E S

T3 T3
(U <u
X X

B a

'g '2 ^ ^

fc -t:; -t;

S S ^ ^

T3

OJ

X

B

<u

'O T3 ^
1.2.2
I EES

X

S S

S H H H s 6 a H

^ ^ ^ ^
o o o o

T3T3x>>'^§§
(u <u M ^ a (u .2 .2

a a'x— ^ a’^'^

a - -

H n— 1 '7-i

o o

a a

a^

73 73
<u (U

a a

o o

I

(u .2

a ^
^ a
^ a

o

a c3

DO O

a .a .a -a

ai^xx

auo'S.a'iH.aiS

H

.a <u

t/5 C/3

o c
o o

lnxt^ooo^O’-|(N

<L>

T3

C

C3

^ s

O

a a

c3

^!U X
D- C
C3 03

m-^mxt^aooNO

aguas fundas brancas low/medium white A- 1, 2, 3, 4, 5 F

aguas fundas pretas low/medium/high black A- 1, 2, 3, 4, 5 F

barreiras low/medium/high white T-1, 2, 3, 4, 5; B, F,

Table 9. — Continued.

1983

Parker et al. — Amazonian Ethnoecology

191

1

^ S
o

^ O

^ ^ ^ ^ (5 ^ ^

pH hP' [C hh'' Hh'' Ph HH h-T Ph HH PP hH ^ Ph Hh"' HH pH

pq pq' pq'' pq" pq" pq" pq’' pq" pp' pq" pq"' Pn" PQ 03 03 03“' 03'' PO' 03''

< < < < < < < < < < < cq' < < < < < < <

m

<N fN (N

— T m — T ir^ m C'j —T

^ ^ ^ ^ ^ I ^ I

fT (N (T cnT rl cT ri (N (T (T bf

^ m

I

<

A

rn ro ^

(N (N (N <N

<U

^ PP ’>
•tP o ^

^ S o
i3
3

^ aj

^ ^ o .pp ^

^ ^ -S -3 ^ ^

o o £ ^ ^ u

^ in

2 S

p^ <u
o .tP
^ PO
-O ^

see
.2 .2 .2
'-B '-B ''B

(U (L» (U

2 2 2

^

o o o

2 2 2

£ £ -PP

G G . ^ ^ ^

„ „ <U (U (U

'^'0^222

(U G \ ^ ^

•2 .2 ^

XS X!

(U G

2 2 ^ ^

o o

§).?
o3 G

2

.2 ^
T3 ^

3

3

><

3

0

X

03

P2

"2

13

"3

3

13

X)

3

3

c

13

N

0

G

G

2

'2

G

3

X3

G

3

03

N

N

■c

13

3

X)

13

N

'2

"2

.2

'3

G

3

G

3

c/l

X!

X!

Xi

T'

o3

• 2

3

•p

o

o

'2

0

3

'g

13

3

3

2

G

3

3

3

3

O

CJ

3

3

a

3

a

3

’3

X)

3

3

3

3

3

G

0

G

0

(U

X>

3

X)

3

G

G

G

2

G

X)

X)

(N

3-

m

VO

OO

os

0

3-

(N

(N

<N

fN

<N

rs

(N

fN

(N

m

m

<3

m

f3

9 o

rv O O
3 TP X3

HH cij 03
ca 0> O'

2 2 2

N O
(U

S=^ C

^ .2

2? a

See Fig. 6.

Letters represent the following: A = animals (these are generally mammals); B = birds; F = fish; I = insects; Rt = reptiles/turtles.

192

Annals of Carnegie Museum

VOL. 52

1983

Parker et al.~ Amazonian Ethnoecology

193

The Caboclo Community of Limoeiro do Ajuru:

Resource Zones and Exploitation
Technologies

Limoeiro do Ajuru is located amidst a broad reach of vdrzea (flood-
plain) lands in the estuary region of Amazonia (Fig. 1). Unlike the
caboclo community of Coari, where seasonal change in water levels is
the key parameter of the local ecosystem, the most important feature
in the Limoeiro do Ajuru area is the fluctuation in daily river levels
(which averages about 3.4 m throughout the year). Indeed, the annual
oscillation in maximum monthly river stages recorded over a 3-year
period was 1 m or less (Fig. 7). Hence, and surprising to many observers,
there is no high-water season of any consequence in the Limoeiro do
Ajuru area— a characteristic that distinguishes the vdrzea of this area
from other vdrzea zones found in Amazonia. Despite the fact that
Limoeiro do Ajuru is situated more than 240 km from the mouth of
the Amazon River, the change in daily river levels (and the flatness of
the annual river stage profile) is due to the encounter between tidal
forces and the enormous volume of water discharged by the Amazon
River— average annual discharge is estimated to be 175,000 m^s“^
(Oltman, 1967). The daily rise from 3 to 4 m in local watercourses
results in the flooding of the dense rainforest (tidal vdrzea forest) that
covers the surrounding landscape (there are no vdrzea lakes or floodable
grasslands in the local area).

The regular inundation of the land limits the opportunity for, and
the importance of, agriculture in the Limoeiro do Ajuru area. In con-
trast to the Coari area where terra firme is abundantly interspersed
throughout the vdrzea, there are precious few pieces of land that are
above daily flood levels and the majority of these parcels are subject
to occasional inundation during peak flood periods. This poses a major
problem for local residents insofar as the principal cultigen of the area,
manioc {Manihot esculenta), is a root crop highly vulnerable to wa-
terlogging. As a consequence of these conditions, and though most
households have small rogas (gardens), the exploitation of aquatic re-
sources represents the most important component of the subsistence
economy, and of local resource perceptions. As such, and given the
scope of this paper, the discussion will focus upon this feature of the
subsistence economy.

The fluvial dynamics of the estuarine vdrzea, and the location of the

Fig. 6. — Schematic of aquatic and terrestrial/arboreal vertical levels from an Amazon
River vdrzea ecological zone near Coari.

194

Annals of Carnegie Museum

VOL. 52

Sia}8i/M uj Hi6!8H

Fig. 7. — River stages at Cameta, Para (55 km upriver from study area), June 1968-May 1971.

1983

Parker et al.— Amazonian Ethnoecology

195

village itself, provide the caboclo of the Limoeiro do Ajuru area with
several microenvironments to exploit. Eight major aquatic resource
zones are identified by residents: the Tocantins River mouthbay, To-
cantins River shore areas, Limoeiro River, igarape (stream) mouths,
large igarapes, small igarapes, flooded forests, and temporary summer
pools (Fig. 8). Within zones, resource perceptions are further articulated
on the basis of ecological variations. These ecological/resource zone
subcategories are defined in terms of water depth, water temperature,
distance from the river bank, type of river bank vegetation, time of
day, meanders, point bars, mudflats, extended shallows, and the level
and velocity of streamflow. The latter two are of importance because
local watercourses have their streamflow direction reversed four times
daily as well as experiencing two distinct high-water (preamar) and two
low-water (baixamar) periods each day. In addition to the above fac-
tors, the time of year is also an important component of aquatic re-
source perception as many species of fish and aquatic animals are
available only during specific periods of the year.

The caboclos of Limoeiro do Ajuru associate particular fish species
with specific aquatic zones. For example, card, jiju {Hoplerythrinus
unitaeniatus), trdira (Hop lias malavaricus), pira coroca, tambaqm (Co-
lossoma macropomum), acari, and piranha (Serrasalmus spp.), are
linked with flooded forest zones as well as with the small igarapes of
the centra (interior forest locales). Indeed, fish feed on the fruits that
fall into the flooded forest and fishing strategies of local residents are
often tied to the fruiting periods of particular species— tambaqm (see
above), a favorite of the local caboclo populace, is known to feed on
the fruits of maraja (Pyrenoglyphis maraja) and seringa (Hevea bras-
iliensis) among others. Dourado (Brachyplatystoma flavicans), mapard
(Hypophthalmus spp.), and pacu are typically found in the open waters
of the Tocantins River mouthbay. Species commonly identified with
igarapes are itui, parapitinga (Colossoma bidens ?), and arucu branco
(Leporinus trifasciatus). Tucunare (Cichla spp.) are common in To-
cantins River shore areas and along the edge of rivers. As would be
expected, the greatest number of fish species are associated with the
main rivers of the local area (such as the Limoeiro River), and include
branquinha (Gasterotomus latior), curimata (Prochilodus nigricans),
pirapucu, and pescada (Palgioscon squamosissimus). Freshwater shrimp,
camardo (Macrobrachium amazonicum), are found in abundance along
the bottom of the large igarapes and rivers of the area.

Other faunal species are linked with specific aquatic resource zones.
Caiman, for example Caiman crocodilus and Paleosuchus trigonatus,
are most often found in large and small igarapes as well as the flooded
forest. Tracaja (Podocnemis unifilis) is found in the flooded forest and
pitiu (P. sextuberculata) in both the flooded forest and shallow areas

196

Annals of Carnegie Museum

VOL. 52

LEGEND:

Tocantins mouthbay
Tocantins shallows
Large rivers (e.g. Limoeiro River)
Large igarapes (streams)

Small igarapes (stream-creek)

Igarape mouth

Flooded forest zones
Temporary pond areas
[ j Floating vegetation zones
Dense forest

Fig. 8. — Aquatic resource zones in the proximity of Limoeiro do Ajuru,

1983

Parker et al. — Amazonian Ethnoecology

197

Table \0.— Aquatic resource zones and methods of exploitation at Limoeiro do Ajuru,

Brazil.

Methods of exploitation

Aquatic resource zones

1 2

3

4

5

6 7

8

9

10 11

Tocantins Mouthbay

X

Tocantins Shore Area

X

X

Limoeiro River

X

X

X

X

X

X

Igarape Mouths

Large Igarape

X

X

X

X

X

X

Small Igarape

X

X

X X

X

X

Hooded Forest
Temporary Summer

X

X

X

Ponds

X

1 . Pari (fish weir).

2. Cacuri (permanent barrier trap).

3. Matapi (cylindrical shrimp trap).

4. Espinhel (trotline).

5. Espeque/Zagaia (pronged fish spear).

6. Arco e Elecha (bow and arrow).

7. Timbo (fish poisons).

8. Tarefa (circular throwing net).

9. Canico (fishing pole/line and hook).

10. Malhadeira (gill net).

1 1 . Gapoia (temporary summer ponds).

of the Tocantins River mouthbay. Jabot i (Testudo tabulata) is iden-
tified with temporary summer ponds and small terra firme patches
within the flooded vdrzea forest. Manatee {Trichecus ininguis), though
rare, are found in the general area and are most often associated with
large igarapes. Two freshwater dolphins, Inia geoffrensis and Sot alia
Jluviatalis, are common but not eaten by local residents because dol-
phins are believed to possess magical powers (Wagley, 1953:237-241).

Specific techniques and instruments are identified with reference to
the exploitation of the various resource zones. These include the pari
(fish weir), cacuri (permanent barrier trap), matapi (cylindrical shrimp
trap), caniqo (fishing pole), espeque/zagaia (pronged fish spear), arco e
flecha (bow and arrow), timbo (fish poisons), tarefa (circular throwing
net), espinhel (trotline), gapoia (drainage of summer ponds), and in
recent years, the malhadeira (gill net). Table 10 shows the techniques/
instruments that are associated with particular aquatic resource zones
(see Parker, 1981:153-159, 274-285, for further discussion).

Two of the fishing instruments, the pari and the cacuri, deserve
special mention because they represent specific instruments adapted
for use in the vdrzea of the Limoeiro do Ajuru area where water levels
fluctuate significantly each day. To construct a pari, a barrier made

198

Annals of Carnegie Museum

VOL. 52

from lengths of split bamboo woven together with strips of vine is
placed across the mouth of a small igarape (stream). The barrier wall
has a gate in the center that is left open as water levels begin to rise
and fish move from the larger watercourses into smaller ones. Just
before the water begins to recede, and the small streams empty, the
gate is closed and the returning fish are thus trapped against the barrier.
The fish swim into a central chamber area where they are easily re-
moved by fishermen. The pari is elegant in its simplicity, temporary,
and can be erected and taken down by a single individual.

The cacuri is designed with the same strategy in mind but has two
important differences— it is permanent, and it is used in the larger
watercourses such as the Limoeiro River (150-200 m wide) and along
the Tocantins River shore area where the pari would be useless. The
cacuri is a barrier trap consisting of a wall of tall stakes sunk into the
river bottom and secured by vines for stability. The wall runs parallel
to the river bank (or shore area) from about 5 to 20 m out into the
water, depending on water depth. As the water level begins to drop,
fish move away from bank areas in search of the security of deeper
water. They encounter the barrier wall and are forced to swim along
it until they are guided into a circular trap located at the center of the
cacuri (both ends of the cacuri are set closer to the shore than the
middle thus encouraging the fish toward the center). The circular trap
is then secured and the fish retrieved.

Confronted with an environment that severely limits opportunities
for agriculture, the caboclos of Limoeiro do Ajuru have evolved an
elaborate matrix for the exploitation of aquatic resources. Within this
matrix, aquatic resource zones and subzones, fish and aquatic animal
species, time of year, and instrument/techniques are linked together to
ensure an adequate source of protein year-round. This knowledge of
aquatic resources, only briefly suggested here, is but one example of
the knowledge possessed by the residents of this area of local environ-
mental variation and nuance. Limoeiro do Ajuru caboclos hunt a wide
range of game animals in the dry forest areas, employ myriad wild
plant species for food and medicinal purposes, and have an elaborate
classification system of woods, their specific properties and appropriate
uses (Parker, 1981:296), and recognize and describe the characteristics
of numerous insects found in the flooded forest— the list goes on. Such
comprehensive ecological knowledge represents a significant source of
data about the variation and richness of Amazon habitats.

Summary

This paper has, albeit in brief fashion, surveyed four Amazon pop-
ulations—two Amerindian societies and two caboclo communities—
with regard to perceptions of specific ecological variations as well as

1983

Parker et al.— Amazonian Ethnoecology

199

resource utilization within their local areas. Long-term resource man-
agement strategies and the semi-domestication of plants and animals
(Kayapo), the selection and configuration of garden sites based upon
edaphic- vegetative-climatic associations (Yekuana), the elaboration of
complex resource unit typologies and vertical zonation within resource
zones (Coari), and vdrzea resource zones and related exploitation tech-
nologies (Limoeiro do Ajuru), are illustrative of the extraordinary
knowledge that peoples indigenous to the Amazon region possess. A
significant range of habitat, ecosystems, and resources are suggested by
these four examples alone — the fact that the foregoing is but a meager
representation of the ethnoecological data and information collected
from these four populations, and that there remain so many groups
and communities yet to be studied in a multitude of ecological settings,
should convince the reader of the enormous potential that folk knowl-
edge systems in Amazonia pose for scientific understanding of the
region and the elaboration of rational, sustainable development strat-
egies.

Development of the Amazon region has proceeded, and will continue
to advance, at a rapid rate. Researchers of every stripe have been
scrambling to keep abreast of development strategies in an effort not
only to provide detailed information for those individuals and agencies
formulating policy in the region but also to identify and classify as
much of the rapidly shrinking Amazonian biome as is possible. While
anthropologists and cultural geographers have continued to study hu-
man populations in the Amazon Basin and their adaptive solutions to
various environmental and ecological milieus, the results and lessons
of these investigations have been largely overlooked, or ignored, by
the larger scientific community and developmental theorists. As recent
events have irrefutably demonstrated, rational development of the Am-
azon region will demand an intimate knowledge and understanding of
the structure and dynamics of ecosystems as well as the complex het-
erogeneity that characterizes the region as a whole.

There is a sad irony in the realization that those who could most
assist in our understanding of this remarkable region are rapidly dis-
appearing as a consequence of modern development— Amerindian and
caboclo knowledge systems are being lost to humanity at an alarming
rate. There is thus a great urgency for the collection of ethnoecological
information from as many diverse social groups, geographic areas, and
ecological microzones in Amazonia as possible. It is extraordinary that
caboclos, who are considered by some to be the dominant cultural force
in Amazonia (Moran, 1974; Ross, 1978), have remained largely ignored
by researchers. To date, there have been only two in-depth studies of
caboclos in the Brazilian Amazon (Wagley, 1953; Parker, 1981). As
caboclos are the principal inhabitants of rural Amazonia, and for all

200

Annals of Carnegie Museum

VOL. 52

practical purposes the sole inhabitants of vdrzea environments, their
knowledge and understanding of the various microenvironments they
oecupy represents a significant source of ethnoecological information.

The question is frequently heard: “What can be done about the
destruetion of Amazonia?” The answer may in part be linked to the
development of an appreciation for the biological and resource diver-
sity of the region as perceived by native peoples. Building upon this
foundation, we can thus begin to elaborate development policies that
are more soeially and environmentally sound. There is a certain hu-
mility with which Western seience must, most uncustomarily, en-
counter folk science in order to benefit from these rich systems of
knowledge. The study of ethnoecology (including ethnobotany and eth-
nozoology, among others), provides a pathway to a productive linkage
between these two systems of knowledge as, hopefully, this paper sug-
gests. There exists enormous potential in ethnoecological inquiry for
increasing our understanding of the natural world, and it is our hope
that in the future there can be a fruitful collaboration between natural
seientists and ethnoecologists in Amazonia as well as other areas of
the world.

Acknowledgments

We wish to thank William Denevan, William Smole, and Hugh Genoways for critically
reading the manuscript. Figures were prepared by the staff of the Cartographic Production
Lab, Department of Geography, University of Maryland Baltimore County.

Literature Cited

Barbira-Scazzocchio, F. (ed.). 1980. Land, people and planning in contemporary
Amazonia. Cambridge Univ. Press, Cambridge, 313 pp.

Beckerman, S. 1980. Fishing and hunting by the Bari of Colombia. Pp. 68-109, in
Studies in hunting and fishing in the Neotropics (R. Hames, ed.), Bennington College,
Bennington, Vermont, 137 pp.

Brokensha, D., D. Warren, and O. Werner (eds.). 1980. Indigenous knowledge

systems and development. Univ. Press America, Latham, Maryland, 474 pp.
Carneiro, R. 1978. The knowledge and use of rain forest trees by the Kuikuru Indians
of central Brazil. Pp. 201-216, in The nature and state of ethnobotany (R. Ford,
ed.). The Anthropological Papers, 67, Museum of Anthropology, Univ. Michigan,
Ann Arbor, Michigan, 428 pp.

Chernela, J. 1982. Indigenous fish and forest management. Cultural Survival, 6:
17-18.

CNBB-CEP. 1 976. Pastoral da terra: posse e conflitos. Estudos da Conferencia nacional
dos Bispos do Brasil— Commissao Episcopal de Pastoral, Paulinas, Sao Paulo,
231 pp.

Davis, S. 1977. Victims of the miracle. Cambridge Univ. Press, New York, 205 pp.
De Carmargo, F. 1949. Reclamation of the “igapos” of Belem, State of Para, Brazil.
Experience paper, section meeting Land Resources 16, United Nations Conference
on the Conservation and Utilization of Resources, New York, 12 pp.

Denevan, W. 1966. A cultural ecological view of the former aboriginal settlement in
the Amazon region. Professional Geographer, 18:346-351.

. 1976. The aboriginal population of Amazonia. Pp. 205-234, in The native

1983

Parker et al.— Amazonian Ethnoecology

201

population of the Americas in 1492 (W. Denevan, ed.), Univ. Wisconsin Press,
Madison, 376 pp.

1982. Ecological heterogeneity and horizontal zonation of agriculture in Ama-
zonia. Paper presented at the Conference on Frontier Expansion in Amazonia, Univ.
Florida, Gainesville, February, 1982.

Denevan, W., J. Treacy, and J. Alcorn. 1 982. Indigenous agroforestry in the Peruvian
Amazon: the example of Bora Indian utilization of swidden fallows. Paper presented
in session on Change in the Amazon Basin, 44th Congress of Americanists, Man-
chester, England, September, 1982.

Fittkau, E., U. Irmler, W. Junk, F. Reis, and G. Schmidt. 1975. Productivity,
biomass, and population dynamics in Amazonia water bodies. Pp. 289-31 1, in
Tropical ecological systems: trends in terrestrial and aquatic water bodies (F. Golley
and F. Medina, eds.). Springer- Verlag, New York, 398 pp.

Frechione, j. 1981. Economic self-development by the Yekuana Amerinds in southern
Venezuela. Unpublished Ph.D. disser., Univ. Pittsburgh, Pittsburgh, Pennsylvania,
351 pp.

. 1983. Manioc monozoning in Yekuana agriculture, (manuscript)

Fuchs, H. 1964. El sistema de cultivo de las Deukohuana (Maquiritare) del alto Rio
Ventuari, Territorio Federal Amazonas, Venezuela. American Indigena, 24:171-
195.

Gentry, A., and J. Lopez-Parodi. 1980. Deforestation and increased flooding of the
upper Amazon. Science, 210:1354-1356.

Gibbs, R. 1967. The geochemistry of the Amazon River system: Part 1. The factors
that control the salinity and the composition and concentration of the suspended
solids. Bull. Geol. Soc. Amer., 78:1203-1232.

Gottlieb, O. 1981. New and underutilized plants in the Americans: solution to prob-
lems of inventory through systematics. Interciencia, 6:22-29.

Goodland, R. 1975. The Cerrado ecosystem of Brazil: state of knowledge report on
tropical ecosystems. UNESCO, Div. of Ecological Science, Paris, 103 pp.

Goodland, R., and H. Irwin. 1975. Amazon jungle: green hell to red desert? Elsevier,
Amsterdam, 155 pp.

Hames, R. 1980. Monoculture, polyculture, and polyvariety in tropical forest swidden
cultivation. Paper presented at American Anthropological Association annual meet-
ings, Washington, D.C., December, 1980.

Hecht, S. 1981. Deforestation in the Amazon Basin: magnitude, dynamics, and soil
resource effects. Studies in Third World Societies, 13:61-100.

— . 1982. Cattle ranching in the eastern Amazon Basin: evaluation of a development

strategy. Unpublished Ph.D. disser., Univ. California, Berkeley, 455 pp.

Indigena. 1974. SUPYSAUA: a documentary report of the conditions of Indian peoples
in Brazil. Indigena and American Friends of Brazil, Berkeley, California, 63 pp.

Irmler, U. 1977. Inundation-forest types in the vicinity of Manaus. Pp. 17-29, in
Ecosystems research in South America (P. Muller, ed.). Junk, The Hague, 366 pp.

Klee, G. (ed.). 1980. World systems of traditional resource management. Halstead,
New York, 290 pp.

Lathrap, D. 1970. The upper Amazon. Praeger, New York, 256 pp.

Lima, R. 1956. A agricultura nas varzeas do estuario do Amazonas. Boletim Technica
do Instituto Agronomico do Norte, Belem, 33:1-64.

Lovejoy, T., and H. Schubart. 1981. The ecology of Amazonian development. Pp.
21-26, in Land, people and planning in contemporary Amazonia (F. Barbira-Scaz-
zocchio, ed.), Cambridge Univ. Press, Cambridge, 313 pp.

Mahar, D. 1979. Frontier development policies in Brazil. Praeger, New York,

182 pp.

Meggers, B. 1971. Amazonia: man and culture in a counterfeit paradise. Aldine,
Chicago, 182 pp.

202

Annals of Carnegie Museum

VOL. 52

Moran, E. 1974. The adaptive system of the Amazonian Caboclo. Pp. 136-159, in
Man in the Amazon (C. Wagley, ed.), Univ. Presses Florida, Gainesville, 330 pp.

. 1981. Developing the Amazon. Indiana Univ. Press, Bloomington, 292 pp.

Oltman, R. 1967. Reeonnaissance investigations of the discharge and water quality
of the Amazon. Pp. 163-185, in Atas do simposio sobre a biota Amazonica. 3.
Limnologia (H. Lent, ed.), Conselho Nacional de Pesquisas, Rio de Janeiro, 325 pp.

Pandolfo, C. 1978. A floresta Amazonica Brasiliera— enfoque economico-ecologico.
SUDAM, Belem, 118 pp.

Parker, E. 1981. Cultural ecology and change; a caboclo varzea community in the
Brazilian Amazon. Unpublished disser., Univ. Colorado, Boulder, 452 pp.

. 1983. Agriculture, development, and the Amazon varzea: a reappraisal.

(manuscript)

Pinto, L. 1977. Amazonia: o anteato da destruigao. Grafisa, Belem, 372 pp.

PiRES, J. 1974. Tipos de vegetasao da Amazonia. Brasil Florestal, 5:48-58.

Posey, D. 1979. Ethnoentomology of the Gorotire Kayapo of central Brazil. Unpub-
lished disser., Univ. Georgia, Athens, 187 pp.

. 1981. Ethnoentomology of the Kayapo Indians of central Brazil. J. Ethno-

biology, 1:165-174.

. 1982. The keepers of the forest. Garden Magazine, New York Botanical Gar-
dens, 6:18-24.

. 1983. Indigenous ecological knowledge and development. Pp. 225-257, in The

dilemma of Amazonian development (E. Moran, ed.), Westview Press, Boulder,
Colorado, 347 pp.

Posey, D., J. Frechione, and L. da Silva. 1983. Folk perceptions of ecological zones
and natural resources in the Brazilian Amazon; ethnoecology of Lake Coari. Human
Ecology (forthcoming).

Prance, G. 1979. Notes on the vegetation of Amazonia III. The terminology of Am-
azonian forest types subject to inundation. Brittonia, 31:26-38.

Rebelo, D. 1973. Transamazonica: integragao em marcha. Ministerio dos Transportes,
Centro de Documentagao e Publica^ao, Rio de Janeiro, 241 pp.

Reichel-Dolmatoff, G. 1976. Cosmology as ecological analysis: a view from the rain
forest. Man, 11:307-318.

. 1978. Desana animal categories, food restrictions, and the concept of color

energies. J. American Lore, 4:242-291.

Ribeiro, D. 1970. Os indios e a civilizagao. Editora Civilizagao Brasiliera, Rio de
Janeiro, 509 pp.

Ross, E. 1978. The evolution of the Amazon peasantry. J. Latin American Studies,
10:193-218.

ScHMiTHUSEN, F. 1978. Coutratos de utilizagao florestal com referenda especial a
Amazonica Brasiliera. IBDF, Brasilia, 33 pp.

ScHUBART, H. 1977. Ecological criteria for the agricultural development of the drylands
in Amazonia. Acta Amazonica, 7:559-567.

SiOLi, H. 1967. Studies in Amazonian waters. Pp. 9-50, in Atas do simposio a biota
Amazonia. 3. Limnologia (H. Lent, ed.), Conselho Nacional de Pesquisas, Rio de
Janeiro, 325 pp.

. 1968. Hydrochemistry and geology in the Brazilian Amazon region. Amazon-

iana, 1:267-277.

— . 1973. Recent human activities in the Brazilian Amazon region and their eco-
logical eflfects. Pp. 321-324, in Tropical ecosystems in Africa and South America:
a comparative review (B. Meggers, E. Ayensu, and W. Ducksworth, eds.), Smith-
sonian Institution, Washington, D.C., 350 pp.

. \915a. Amazon tributaries and drainage basins. Pp. 199-213, in Coupling of

land and water systems (A. Hasler, ed.). Springer- Verlag, New York, 336 pp.

1983

Parker et al.~ Amazonian Ethnoecology

203

. \915b. Tropical river: The Amazon. Pp. 461-488, in River ecology (B. Whitten,

ed.), Blackwells, Oxford, England, 560 pp.

Smith, N. 1981. Man, fishes, and the Amazon. Columbia Univ. Press, New York,

180 pp.

Smole, W. 1976. The Yanomano Indians: a cultural geography. Univ. Texas Press,
Austin, 286 pp.

SoMBROEK, W. 1966. Amazon soils: a reconnaissance of the soils of the Brazilian
Amazon region. Centre for Agricultural Publications and Documentation, Wage-
ningen. The Netherlands, 292 pp.

Sternberg, H. 1975. The Amazon River of Brazil. Steiner-Verlag, Wiesbaden, West
Germany, 74 pp.

Steward, J. (ed.). 1 946-59. Handbook of South American Indians. Bureau of American
Ethnology, Washington, D.C., 1:1-624 (1946); 2:1-1035 (1946); 3:1-986 (1948); 4:
1-609 (1948); 5:1-818 (1949); 6:1-715 (1950); 7:1-286 (1959).

UNESCO. 1978. Tropical forest ecosystems: a state-of-knowledge report. UNESCO,
Paris, 683 pp.

Velho, O. 1972. Frentes de expansao e estrutura agraria: estudo do processo de pe-
netracao numa area da Transamazonia. Zahar, Rio de Janeiro, 178 pp.

Vickers, W. 1979. Native Amazonian subsistence in diverse habitats: the Siona-Secoya
of Ecuador. Studies in Third World Societies, 7:6-36.

Wagley, C. 1953. Amazon town. Macmillan, New York, 336 pp.

WissLER, C. 1917. The American Indian. McCurtire, New York, 435 pp.

Wood, C., and M. Schmink. 1979. Blaming the victim. Studies in Third World So-
cieties, 7:77-93.

Back issues of many Annals of Carnegie Museum articles are
available, and a few early complete volumes and parts are listed
at half price. Orders and inquiries should be addressed to: Pub-
lications Secretary, Carnegie Museum, 4400 Forbes Avenue,
Pittsburgh, Pa. 15213.

I

I

r"

" hE!^ -

\

y

I

,/

V

?■>

K

V

7. 73

ISSN 0097-4463

AN NALS

o; CARNEGIE MUSEUM

CARNEGIE MUSEUM OF NATURAL HISTORY

4400 FORBES AVENUE * PITTSBURGH, PENNSYLVANIA 15213

VOLUME 52 16 SEPTEMBER 1983 ARTICLE 9

REVISION OF THE WIND RIVER FAUNAS, EARLY
EOCENE OF CENTRAL WYOMING.

PART 3. MARSUPIALIA

Leonard Krishtalka
Associate Curator, Section of Vertebrate Fossils

Richard K. Stucky
Postdoctoral Fellow, Section of Vertebrate Fossils

Abstract

Dental remains of three species of Peratherium {P. comstocki, P. marsupium, and P.
innominatum), one species of Peradectes {P. chesteri) and two species in a new genus,
Armintodelphys (A. blacki, A. dawsoni) are reported from the Wind River Formation,
north-central Wyoming. This is the first Wasatchian record of P. marsupium and, with
P. innominatum, the first record of marsupials from the Lysite Member. The range of
P. comstocki is extended into the early Bridgerian. Review of other Eocene Didelphini
indicates that: Entornacodon minutus and Peratherium morrisi belong in P. knightp P.
innominatum is a valid species of Peratherium-, P. macgrewi is a subspecies of P. in-
nominatum-, and P. knighti occurs in Wasatchian deposits. Among Peradectini, Pera-
dectes chesteri is distinct from P. protinnominatus and is morphologically and phylo-
genetically intermediate in a lineage that includes the latter and P. minutus {=Nanodelphys
minutus)-, Armintodelphys is most closely related to P. pauli.

Introduction

Almost one hundred years ago Cope (1884:269) described Pera-
therium comstocki from the “badlands of the Wind River, Wyoming.”
Since then only three specimens have been added to the marsupial
record from the Wind River Formation (Guthrie, 1971), all from the
Lost Cabin Member— two were described as Peratherium cf P. chesteri
and one as P. comstocki. Subsequent work by field parties from the
Carnegie Museum of Natural History and the University of Colorado

Submitted 18 January 1983.

205

206

Annals of Carnegie Museum

VOL. 52

Museum (see Stucky and Krishtalka, 1982) has yielded 47 specimens
of didelphids that are referred to three genera and six species. Three
are known species of Peratherium (P. comstocki\ P. marsupium Troxell,
1923; P. innominatum Simpson, 1928), one belongs to Peradectes (P.
chesteri) and two are new species in a new genus of Peradectini, Ar-
mintodelphys. Examination of material related to these studies has led
to a review of the occurrences, systematics, and evolutionary relation-
ships of Paleocene and Eocene North American didelphids, which will
appear elsewhere (Krishtalka and Stucky, 19836f).

Abbreviations used in this paper are as follows: AMNH, American Museum of Natural
History; CM, Carnegie Museum of Natural History; PM, Field Museum of Natural
History; PU, Princeton University (Museum); UCM, University of Colorado Museum;
UCMP, University of California Museum of Paleontology; USNM, U.S. National Mu-
seum; UW, University of Wyoming; YPM, Yale Peabody Museum; Fm, Formation;
loc., locality; L, length; W, width.

Localities and Age

The didelphid material from the Wind River Formation was re-
covered by surface prospecting and underwater screening of sediments
from seven localities in the Lost Cabin Member (CM iocs. 34, 1039,
1040, 90, 1085; UCM Iocs. 79040, 80061), two in the Lysite Member
(CM Iocs. 931, 932), and two in a previously unnamed sequence of
strata that is not referrable to either of these members (UCM Iocs.
80062, 81008; Stucky and Krishtalka, 1982). This sequence is here
termed the ‘"Red Creek facies.” One of us (Stucky, 1 982; in preparation)
has proposed that three of the localities in the Lost Cabin Member
(CM loc. 34; UCM Iocs. 79040, 80061) are not Lostcabinian but Gard-
nerbuttean (Robinson, 1966), which is now thought to represent the
earliest part of the Bridgerian, and is older than Bridger A (McGrew
and Sullivan, 1970). Briefly, this conclusion is based on: (1) the re-
striction of Lambdotherium to localities in the Red Creek facies and
lower part of the Lost Cabin Member of the Wind River Formation,
and to localities in other basins in western North America that preserve
Wasatchian faunas; (2) the first appearance of Antiacodon, Trogosus,
Palaeosyops, Hyrachyiis, and other taxa at localities in the upper part
of the Lost Cabin Member; these localities are stratigraphically above
those that record the last appearance of Lamhdotheriuny faunas with
these taxa from other basins in western North America have tradi-
tionally defined the Bridgerian Land Mammal Age. For the same rea-
sons, an early Bridgerian Age is assigned to faunas from the upper part
of the Huerfano Formation (Huerfano B) and the Cathedral Bluffs
tongue of the Wasatch Formation, a conclusion also advocated by West
(1973), West and Dawson (1973), and Gingerich (1979). The Bridgerian
age of taxa from localities in the upper part of the Lost Cabin Member

1983

Krishtalka and Stucky— Wind River Faunas, 3

207

is reflected in the systematics section below. In addition, we recognize
that the Bridger Basin is structurally a part of the Green River Basin.
We continue its use as a geographic reference in deference to historical
tradition and greater geographic precision.

Systematics

Family Didelphidae Gray, 1821

In the most recent review of Tertiary didelphids Crochet (1977, 1979)
recognized five genera: Peratherium Aymard, 1850 (four species, early
Eocene-Oligocene, Europe); Amphiperatherium Filhol, 1879 (four
species, early Eocene-middle Miocene, Europe); Herpetotherium Cope,
1873 (six species, early Eocene-early Miocene, North America); Per-
adectes Matthew and Granger, 1921 (four species, late Paleocene-mid-
dle Eocene, North America; two species, early Eocene, Europe); and
Nanodelphys McGrew, 1937 (two species, middle Eocene-middle Oli-
gocene. North America). Bown and Rose (1979) subsequently named
a new didelphid genus, Mimoperadectes.

The six species of Herpetotherium had long been included in Per-
atherium 1884; Simpson, 1928, 1968; Setoguchi, 1975; Green

and Martin, 1976), but Crochet (1977) resurrected the former on the
basis of two diagnostic features— dominance of stylar cusp D on the
upper molars (as opposed to B) and a less reduced talonid on the last
molar. These features alone do not seem to warrant recognition of
Herpetotherium, and are, in any case, variable within and among the
North American Eocene species he attributed to Herpetotherium. Fox
(manuscript; personal communication, 1982), however, recognizes
Herpetotherium for the type species, H.fugax, based on unique features
of the anterior dentition, as well as the characters outlined by Crochet
(1977, 1979) for the upper and lower molars. Accordingly, Peratherium
is retained here for five North American Eocene species (see below),
at least until H. fugax-\\\iQ anterior dentitions are recovered for these
species. In agreement with Bown (1979), and as discussed elsewhere
(Krishtalka and Stucky, 1983^2), Nanodelphys is a junior synonym of
Peradectes. Thus, recognized genera of North American Tertiary di-
delphids are: Peratherium, Herpetotherium, Peradectes, Mimoperadec-
tes, and a new genus from the Wind River Formation, Armintodelphys.

Until most recently, identification of these genera relied primarily
on features of the upper molars. Lower molars were thought to be of
little diagnostic value until Setoguchi (1973, 1975) and Crochet (1977,
1979) distinguished Peratherium from Peradectes and Nanodelphys on
discrete differences in the structure of the hypoconulid-entoconid com-
plex. Bown (1979) and Rose (1981) also adopted these criteria. Cro-

208

Annals of Carnegie Museum

VOL. 52

chefs (1979) two tribes of didelphines— Didelphini and Peradectini —
can be distinguished by diagnostic characters on their lower molars,
including the structure of the hypoconulid, entoconid, and entoconid
notch, and the comparative size of the trigonid and talonid. Also useful
on the upper molars is the size of the conules, stylar cusp C, and the
paracone and metacone, the presence or absence of dilambdodonty,
and the shape of the posterolingual area of the protocone.

Tribe Didelphini Crochet, 1979

North American Tertiary genera in this tribe are Peratherium and
Herpetotherium. Unlike members of the Peradectini, these genera have
tall, spire-like entoconids, much lower, posteriorly projecting hypo-
conulids, and deep, wide entoconid notches on M,_4. have a pos-
terolingually expanded protocone, strong conules and stylar cusp C,
dilambdodonty, and lower paracone than metacone.

Peratherium Aymard, 1850

Peratherium (and Herpetotherium) are distinct from other genera of
North American Tertiary didelphids in three major features of the
molar dentition. (1) On the paracone is at least half the size and
height of the metacone — often and in part a function of wear (in unworn
specimens the paracone may be only slightly lower than the metacone;
see Fig. 4); (2) the centrocrista is dilambdodont; (3) on Mi_3 the ento-
conid is large, high and conical, whereas the hypoconulid is a much
lower, flat, shelf-like cusp that projects posteriorly behind the ento-
conid. Although the two cusps are twinned, they are separated by a
wide, deep valley (entoconid notch). In the other North American
Tertiary didelphids, all of which belong to the Peradectini (except pos-
sibly Thylacodon pusillus; Krishtalka and Stucky, 1983i2), the paracone
and metacone are more nearly subequal and not dilambdodont; the
entoconid and hypoconulid are smaller, subequal, and more closely
twinned; the two cusps arise from a common internal talonid wall and
are separated by a much weaker entoconid notch. Peratherium lacks
the distinctive anterior dentition of Herpetotherium.

Our analysis has led to the recognition of five Eocene species of
Peratherium. In order of decreasing size they are: P. comstocki Cope,
1884; P. edwardi Gazin, 1952; P. marsupium Troxell, 1923; P. knighti,
McGrew, 1959; and P. innominatum Simpson, 1928.

Peratherium morrisi Gazin, 1962, from the Cathedral Bluffs tongue
of the Green River Formation, southwestern Wyoming, is a junior
synonym of P. knightr, the type of the former is indistinguishable from
lower molars in McGrew’s (1959) hypodigm of P. knighti. Setoguchi
(1973), misled by Gazin’s (1952) figure of P. morrisi, erred when he
suggested that this species belonged in either Peradectes or Nanodel-

1983

Krishtalka and Stucky--Wind River Faunas, 3

209

phys. The type of P. morrisi bears the entoconid-hypoconulid structure
that is characteristic of Peratherium and is close in size to P. knighti.

The type of Entomacodon minutus Marsh, 1872, is, as suggested by
Robinson (1968), a marsupial and is also conspecific with Peratherium
knighti. Peratherium has priority over Entomacodon, as does E. mi-
nutus over P. knighti, but the resultant P. minutum (Marsh, 1872) is
preoccupied by P. minutum (Aymard, 1846). In sum, P. knighti in-
cludes E. minutus and P. morrisi. We have identified two teeth (Mi,
CM 42139; M2 or M3, UCMP 59131) of R knighti from the Four Mile
fauna (McKenna, 1960), extending the earliest known record of this
species from the early Bridgerian to the early Wasatchian.

Peratherium chesteri Gazin, 1952, from the La Barge fauna, has been
transfered to Peradectes (Setoguchi, 1973; Bown, 1979), an action with
which we agree.

As discussed in detail below, Peratherium innominatum Simpson,
1928, belongs in Peratherium rather than Peradectes (contra Setoguchi,
1973; Bown, 1982), and P. macgrewi Bown, 1979 is a Graybullian
subspecies of P. innominatum.

Finally, although Setoguchi (1973) and Bown and Rose (1979) ad-
vocated synonymy of Peratherium edwardi Gazin, 1952 with P. com-
stocki, we recommend at least tentative recognition of P. edwardi. The
type, USNM 19200, and a referred specimen, USNM 19206, both
from the La Barge fauna, are, as Gazin indicated, intermediate in size
between P. comstocki and P. marsupium; until a larger sample of this
species is recovered, allocation of P. edwardi to either P. comstocki or
P. marsupium is unwarranted.

Peratherium comstocki CoytQ, 1884
(Fig. 1, 2; Table 1, 2)

AMNH 4252, partial left dentary with M2_3.

Type locality. — \JncQrXdL\n. Cope (1884) reported the type from the
“badlands of the Wind River, Wyoming,” and named the species for
Professor Theodore D. Comstock, a geologist from Cornell University,
who had explored the Wind River region. Matthew (1899:31) listed
the type as Didelphys"' comstocki from the Bighorn Basin, and since
then both basins have been cited as the type locality. Matthew (1909:
92), Osborn ( 1 909:46), Troxell (1923) and Hay ( 1 930) favored the Wind
River Basin; Simpson (1928) and later students, the Bighorn Basin.
Cope (1884) acknowledged his error (Cope, 1880) in describing a Big-
horn Basin collection of Eocene vertebrates as coming from the Wind
River Basin, but it appears that the three type specimens in that col-
lection did not include AMNH 4252 (Gazin, 1953; Gingerich, 1980).
Whatever the answer to this little mystery, P. comstocki has since been

210

Annals of Carnegie Museum

VOL. 52

Fig. l.-Peratherium cornstocki. (A) CM 55562, LM,_2; (B) CM 21126, LM3^; both
approx. X 15.

1983

Krishtalka and Stucky— Wind River Faunas, 3

211

Fig. 2.-Peratherium comstocki. (A) CM 35842, LM'; (B) UCM 44573, RM"; both
approx. X 14.

recovered from both the Bighorn and Wind River basins (Guthrie,
1971; Schankler, 1980; this paper).

Referred specimens. 40078, 55567; M, 2-CM 55562; M3^-CM 21126;

M,-CM 55561; M4-CM 40079; M'-CM 35842; M^-UCM 44573; upper molar
frag., UCM 45337.

Localities.— CM loc. 34 (Gardnerbuttean, Lost Cabin Member); UCM
loc. 80062 (Lostcabinian, Red Creek facies); both in the Wind
River Basin (Wind River Fm.), Wyoming.

Known distribution.— — River Basin (Wind River

212

Annals of Carnegie Museum

VOL. 52

Table Dimensions of lower molars o/Peratherium and Armintodelphys, new genus,
from the Wind River Formation, northcentral Wyoming.

Museum and
catalog no.

Locality

M, M2

M3

M4

L W L

w

L

w

L

w

Peratherium comstocki

CM 55561

34

2.7 1.6

CM 55562

34

2.7 1.6

CM 55567

34

3.1

2.0

CM 55568

34

3.1

2.2

CM 40078

34

3.0

2.0

3.0

2.0

CM 21126

34

2.0

3.0

1.4

CM 40079

34

2.9

1.4

Peratherium rnarsupium

CM 41157

34

1.3

UCM 45234

79040

2.7

1.5

CM 37099

931

2.6

1.6

2.7

1.6

1.2

UCM 42869

79040

2.9

1.7

Peratherium innominatum

UCM 44576

80062

1.6 0.8

UCM 42864

79040

1.8 1.0

UCM 44866

80061

1.7

1.0

UCM 42870

79040

1.7

1.0

UCM 42865

79040

1.9

1.1

CM 22001

90

1.7

1.0

CM 36531

34

1.7

1.1

Armintodelphys blacki

CM 41159 (Type) 1039

2.2

1.1

2.2

0.9

CM 41161

1040

1.0

UCM 45252

80061

2.0

1.1

Armintodelphys dawsoni

CM 55560

34

1.5

0.9

CM 55569 (Type) 34

0.8

1.5

0.8

Fm.), Bighorn Basin (Willwood Fm.), Wyoming; ?San Juan Basin (San
Jose Fm., Lucas et al., 1981), New Mexico. Earliest Bridgerian — Wind
River Basin (Wind River Fm.), Wyoming; Huerfano Basin, (Huerfano
Fm.), Colorado {P. cf. comstocki, Simpson, 1968). Late Bridgerian —
Bridger Basin (Bridger Fm.), Wyoming. Late Bridgerian or early Uin-
tan— Agua Fria (lower Buck Hill Group; West, 1982), Texas.

Emended diagnosis.— EdiXgQsX known North American Eocene species
of Peratherium; with greatly enlarged stylar cusps; labial expan-
sion of stylar cusps B and D create shallow ectoflexus; cusp C medial,
equidistant and separate from B and D; lower molars with propor-
tionately longer talonids.

1983

Krishtalka and Stucky— Wind River Faunas, 3

213

Table I.-— Dimensions of upper molars q/’Peratherium /rc>m the Wind River Formation,

north-central Wyoming.

Museum and
catalog no.

Locality

M' M2

M^

M'*

L

W L W

L

w

L

w

Peratherium comstocki

CM 35842

34

3.1

2.8

UCM 44573

80062

2.8 3.3

Peratherium marsupium

CM 41165

1085

2.5

2.6

UCM 44870

80061

2.7

2.5

UCM 44580

80062

2.6

2.7

UCM 44545

79040

2.7

2.9

2.3

2.8

Peratherium innominatum

UCM 44869

80061

1.6

1.8

UCM 42872

80061

1.6

1.6

UCM 45285

80061

1.9 1.7

UCM 44867

80061

1.3

1.9

Description. — On the first three lower molars the entoconid is large, high and conical,
and is isolated from the proximal hypoconulid by a deep, wide notch. The hypoconulid,
flat and shelf-like, juts posteriorly behind the entoconid. The talonid is longer and wider
than the trigonid, the paraconid is the lowest of the trigonid cusps, and the metaconid
is lingual and slightly posterior to the higher protoconid. The precingulid is strong but
short, ending posteriorly at the base of the crown below the apex of the protoconid. The
hypoconid is broader than the entoconid, but the two cusps are subequal in height. The
cristid obliqua is slightly convex labially and meets the trigonid below and labial to the
notch of the protocristid. The postcristid descends posterolingually from the apex of the
hypoconid to the hypoconulid. From the latter a postcingulid extends labially to the
posterior part of the base of the hypoconid.

M, is shorter and narrower than M2 or M3 and its paraconid leans more anterodorsally.
M4 differs from the other lower molars in having a more elongate talonid that is also
narrower than the trigonid, a broader paraconid than metaconid, and in lacking a postcin-
gulid.

The only known M' (CM 35842) bears a large metastylar wing and a weak ectoflexus,
which is formed by the labial expansion of stylar cusps B and D. The protocone occurs
anteriorly, in line with the paracone and stylar cusp B, so that the posterolingual face of
the protocone is expanded and slightly “squared-off.” The paracone and much higher
metacone are subcrescentic and dilambdodont, with a buccally directed centrocrista. The
stylar shelf is worn, but it appears that cusp D, elongate in shape, was dominant in the
unworn condition, followed in size by cusp B and subequal cusps C and A. A ridge,
perhaps developed through wear, joins cusp C to the buccal extension of the centrocrista.

On M^ (UCM 44573) all of the stylar cusps appear hypertrophied. Cusp D, conical
rather than elongate, is slightly larger than cusp B, and both are approximately twice as
large as the subequal C and A cusps. Cusp C is equidistant from cusps B and D and
directly buccal to the apex of the centrocrista. A slight ectoflexus between cusps B and
D is more a function of the labial expansion of these cusps than a true emargination of

214

Annals of Carnegie Museum

VOL. 52

the stylar shelf. Cusps B, C, and D are separated by acute notches, whereas a weak ridge
connects cusps A and B.

Discussion.— T\\q referred lower and upper molars bear the diag-
nostic features of Peratherium described above. These teeth are sig-
nificantly larger than those assigned below and in other studies to either
P. marsupium, P. knighti, or P. innominatum. M2 and M3 agree in size
and structure with those on the type of P. comstocki. The first and
second upper and lower molars referred here were previously unknown
for this species. CM 21 126, cited in Guthrie (1971) as M2_3, preserves
M4 and part of M3.

We agree with Simpson’s (1968) allocation of UCM 26541, from the
upper part of the Huerfano Formation, to Peratherium cf. comstocki.
This specimen, and those from CM loc. 34 are the first records of this
species from the early Bridgerian. CM 1 390 1 , M3_4 from the upper part
of the Bridger Formation, extends the range of P. comstocki into the
late Bridgerian, and the material from Agua Fria allocated to this
species (West, 1982) may be late Bridgerian or early Uintan. Finally,
AMNH 56307, originally described as an M4 of P. comstocki from the
early Eocene of the Powder River Basin (Delson, 1971), was subse-
quently identified as an M2 of Mimoperadectes (Bown and Rose, 1979).

Peratherium marsupium (Troxell, 1923)

(Fig. 3; Table 1, 2)

Type. — YPM 13518, partial right dentary with P3, Mi_3.

Type locality. — BridgQV Basin, Wyoming.

Referred specimens. 37099; M,-CM 41157, 41158; or M3-UCM
42869, 44336, 45234; M4-UCM 46719; M^-^-UCM 45545; M'-CM 41165; M^-
UCM 44580, 44870; DP^-UCM 44574.

Localities. — CM. loc. 931 (Lysitean, Lysite Member); CM loc. 34,
UCM Iocs. 80061, 79040 (Gardnerbuttean, Lost Cabin Member); UCM
loc. 80062 (Lostcabinian, Red Creek facies); CM loc. 1085 (subage
uncertain. Lost Cabin Member); all in the Wind River Basin (Wind
River Fm.), Wyoming.

Known distribution.— Y/diSdiXchmn through Duchesnean — Wind Riv-
er Basin (Wind River Fm.; Wagon Bed Fm.), Wyoming. Bridgerian—
Bridger Basin (Bridger Fm.), Bighorn Basin (Blue Point Member; Eaton,
1982), Wyoming; Uinta Basin (Green River Fm.), Utah (Krishtalka
andStucky, 1983/?). Late Bridgerian or early Uintan — Sand Wash Basin
(Washakie Fm.), Wyoming; Agua Fria (lower Buck Hill Group; West,
1982), Texas.

Emended diagnosis. — CdiVgQr than P. knighti and P. innominatum;
smaller than P. comstocki and P. edwardi; unlike P. comstocki, stylar

1983

Krishtalka and Stucky—Wind River Faunas, 3

215

Fig. 3.—Perathenum marsupium. CM 37099, LM2_3; approx. X 13.

cusps not bulbous labially; unlike P. knighti, ectoflexus deeper on
and especially M^.

Description. — The lower molars closely resemble those of P. comstocki except for two
features also noted by Simpson (1928): smaller size and slightly shorter talonids in
proportion to the trigonids. M‘, like that of P. comstocki, has a large metastylar salient,
and have an anterior protocone that is elongated posterolingually. On the only
referred M'-— a worn specimen— stylar cusps B and D are subequal and small, followed
in size by cusp A and a tiny cusp C. There is no evidence of a cusp E. On cusp A
is absent, cusp C is joined to D, and the metaconule is larger than the paraconule. Unlike
on the paracone is dominant on M"*, cusp A is the only stylar cusp, and the crown

216

Annals of Carnegie Museum

VOL. 52

is quadrate. A stylar ridge extends from posterior to cusp A to the area labial to the
metacone. The DP^ referred here (UCM 44574) lacks a stylar shelf and has a bulbous
metacone.

Discussion. — ThQ recovery of P. marsupium from the Wind River
Formation represents the first Wasatchian record of this species and,
with P. innominatum, the first known occurrence of marsupials in the
Lysite Member. One of the specimens (CM 23194) that West and
Dawson (1975) assigned to P. marsupium is closer in size and referred
to P. knight i.

Peratherium innominatum Simpson, 1928
(Fig. 4; Table 1, 2)

— AMNH 11493, partial left dentary originally preserving

Type /6>ci2//Yy. — Millersville, Lower Bridger, Bridger Basin, Wyo-
ming.

Referred specimens. 44576, 42864, 44606; M2 or M3~-CM 22001, 36531,
36967, UCM 44578, 42865, 42870, 42866; M4-UCM 44575, 44577; M>-UCM 42872,
44869; M^, M^-UCM 46614; M^-UCM 45285; M^ or M^-UCM 44868, CM 22000,
55564; M^’-UCM 44881, 44867, 45255; upper molar fragment, UCM 44579.

Localities. — CM loc. 932 (Lysitean, Lysite Member); UCM Iocs.
80062, 81008 (Lostcabinian, Red Creek facies); CM loc. 90 (?Lostcab-
inian. Lost Cabin Member); CM loc. 34, UCM Iocs. 79040, 80061
(Gardnerbuttean, Lost Cabin Member); all in the Wind River Basin
(Wind River Fm.), Wyoming.

Known i7z.s7r/7?w//(9«. — Wasatchian— Wind River Basin (Wind River
Fm.), Bighorn Basin (Willwood Fm., record of P. macgrewi, see below),
Wyoming; Four Mile area (Wasatch Fm.; see below), Colorado. Brid-
gerian — Wind River Basin (Wind River Fm.), Bridger Basin (Bridger
Fm.), Bighorn Basin (“Aycross-like beds,” Eaton, 1982), Wyoming;
Uinta Basin (Green River Fm.), Utah. Uintan and Duchesnean— Wind
River Basin (Wagon Bed Fm.), Wyoming.

Emended diagnosis. — Much smaller than P. comstocki and P. mar-
supium’, consistently smaller than P. knightf also differs from P. knighti
in having distinct ectoflexus on M^ and especially on M^.

Description.— lihQ lower molars referred here resemble the type, AMNH 1 1493, in
size and in having tall entoconids, low, posteriorly projecting hypoconulids, deep, wide
entoconid notches, and compressed trigonids. The upper molars have much higher meta-
cones than paracones and posterolingually expanded protocones. On M' (UCM 42872)
stylar cusp B is well-developed and equal in size to a twinned C + D cusp, which occurs
labial to the premetacrista. The anterior border of cusp C is directly labial to the dorsal
apex of the centrocrista and separated by a large gap from cusp B. The stylar shelf is
indented by a weak ectoflexus between cusps B and C. Stylar cusp A is extremely weak;
there is no cusp E. On M^ (UCM 45285), B is the largest of the stylar cusps; cusps C
and D— proximal but separate— and A are subequal and larger than E.

1983

Krishtalka and Stucky — Wind River Faunas, 3

217

Fig. 4.—Pemtherium innominatum. (A) CM 36531, RM2 or M3, approx. X 14; (B) UCM
45285, RM2; approx. X 11.

Discussion. — Setoguchi (1973) and Bown (1982) allocated Pera-
therium innominatum to Peradectes. When Simpson (1928) described
the type, AMNH 1 1493, it preserved Mi_4. Unfortunately, M, is now
missing, and the entoconid and protoconid of M3 are broken, appar-
ently due to subsequent attempts at further preparation and/or casting
of the specimen. Nevertheless, the size of the entoconid on M2 and its
broken base on M3 is clearly Peratherium-like. In these diagnostic
features, as well as in the size and morphology of M2_4, AMNH 1 1493
is virtually identical to large samples of Peratherium from Powder
Wash, Uinta Basin, and from the Wind River Formation. These sam-
ples are described here and elsewhere (Krishtalka and Stucky, 1983Z?)
as Peratherium innominatum.

218

Annals of Carnegie Museum

VOL. 52

The two teeth (CM 22000, CM 22001) that Guthrie (1971) assigned
to Peratherium cf. P. chesteri are referred here to P. innominatum. A
lower jaw that West (1973, PL l,Fig. B; PM 1 5320) figured and referred
to P. innominatum may not be a marsupial; two other specimens (West,
1973: PL 1, Fig. C, D; PM 15682, PM 15866) he identified as P.
innominatum are referrable to Peradectes chesteri. UW 984 from Tab-
ernacle Butte, identified as Peratherium cf innominatum by McGrew
(1959), bears the diagnostic features of Peradectes and is referred below
and elsewhere (Krishtalka and Stucky, 1983^2) to P. chesteri. Our anal-
ysis of the Badwater late Eocene didelphids indicates that Setoguchi’s
(1975) sample of Peratherium cf P. knighti includes specimens of P.
knighti and P. innominatum.

P. innominatum is closest in size and morphology to P. knighti and
P. macgrewi. The teeth are smaller than those of P. knighti and M^“^
have more distinct ectoflexi. Distinction of P. innominatum from P.
macgrewi other than on mean size is not possible with the knov/n
material. Except for M,, upper and lower molars of P. innominatum
are slightly but consistently larger. The current stratigraphic record and
the close similarity between the Graybullian P. macgrewi and the Lys-
itean through early Bridgerian P. innominatum, imply an ancestor-
descendant relationship between the two that involved a slight increase
in size. Similar anagenetic increase in size occurs between the Wind
River Formation population of P. innominatum and that from Powder
Wash, Green River Formation, Utah (Krishtalka and Stucky, 1983/?).

P. macgrewi is reduced in rank to a subspecies of P. innominatum for
biostratigraphic purposes (Simpson, 1961:175).

Peratherium innominatum macgrewi (Bown, 1979), new rank

This temporal subspecies of P. innominatum is known only from
Graybullian horizons in the Bighorn Basin and the Sand Wash Basin.
We have identified an isolated M2 or M3 (CM 42137) of P. i. macgrewi
from Sand Quarry, Four Mile fauna, Colorado. Also, as noted by
McKenna (1960, Fig. 18a), UCMP 44095 from Kent Quarry is Per-
atherium-\ike. The disparity in height between the paracone and meta-
cone on M^“^, the dilambdodont centrocrista, and the expanded pos- j
terolingual area of the protocone distinguish this specimen from
contemporaneous Peradectes in the Four Mile sample. The size of
M^"^ on UCMP 44095 is within the range of these teeth in the hypodigm
of P. i. macgrewi from the type area (Bown, 1979). Also, one of the
upper molars in UW 9742 that Bown (1979, Fig. 40A, center) identified ;
as P. chesteri is dilambdodont and belongs to P. i. macgrewi.

Tribe Peradectini Crochet, 1979

North American Tertiary genera in this tribe are Peradectes (includ-
ing Nanodelphys, see Bown, 1979; Krishtalka and Stucky, 1983a),

1983

Krishtalka and Stucky— Wind River Faunas, 3

219

Mimoperadectes, and Armintodelphys, the new genus from the Wind
River Formation. They differ from Peratherium in that Mi_4 bear a
low or very reduced entoconid, a closely twinned and dorsally pro-
jecting hypoconulid, and a weak or vestigial entoconid notch. Addi-
tionally, have a V-shaped protocone (no posterolingual expan-
sion), conules and stylar cusp C that are small to absent, no
dilambdodonty, and a paracone that is as high as or slightly lower than
the metacone.

Pemdectes Matthew and Granger, 1921

Peradectes is much smaller than Mimoperadectes and has a larger
metaconid than paraconid on M2_4. Compared to Armintodelphys, low-
er molars of Peradectes retain subequal hypoconulid and entoconid
and an entoconid notch.

Our revision of the systematics of Paleogene North American di-
delphids (Krishtalka and Stucky, 1983a) indicates that there are six
discernable species of Peradectes: P. elegans Matthew and Granger,
1921; P. pauli Gazin, 1956; P. protinnominatus McKenna, 1960; P.
chesteri (Gazin, 1952); P. californicus (Stock, 1936; see Lillegraven,
1976); P. minutus (McGrew, 1937). Of these, only P. chesteri has been
found in the Wind River Formation. Thylacodon pusillus may also be
a species of Peradectes (Clemens, 1979) and is under study elsewhere
(see Archibald, 1982).

Peradectes chesteri (Gazin, 1952)

— USNM 19199, right partial dentary with M3.

Type locality. — UppQv part of the Wasatch Formation (=“Upper
Knight beds”). La Barge, Wyoming.

Referred specimen. — ~\J CM 45288.

Locality.— \] CM. loc. 80061 (Gardnerbuttean, Lost Cabin Member),
Wind River Basin (Wind River Fm.), Wyoming.

Known distribution. — LslXq Wasatchian— Green River Basin (Wasatch
Fm.), Wyoming. Bridgerian — Wind River Basin (Wind River Fm.),
Green River Basin (Bridger Fm.), Wyoming; Uinta Basin (Green River
Fm.), Utah.

Emended diagnosis. — CompdLVQd to P. elegans: teeth smaller; Mi
smaller than M2; M3 talonid narrower (basally) than trigonid; compared
to P. elegans and P. pauli: P3 lower than Mi and with shorter talonid;
M4 talonid shorter than trigonid; compared to P. elegans, P. pauli and
P. protinnominatus: Mi_3 talonids with narrower occlusal width (hy-
poconid more internal); compared to P. elegans, P. pauli, P. protin-
nominatus and P. californicus: greater disparity in L/W ratio from M*
to M^; conules and stylar cusps vestigial; M^ more transverse with a
more highly compressed protocone; Mi_3 narrower in proportion to

220

Annals of Carnegie Museum

VOL. 52

length, with no labial emargination between trigonid and talonid; com-
pared to P. californicus: P3 talonid present; compared to P. minutus:

paracone higher than protocone and stylar cusp B; stylar cusp B
not enlarged; and especially less transverse, with less compressed
protocone; gap between M4 and ascending ramus present.

Description.— The isolated M‘ (L = 1.2; W = 1.1) is abraded, but preserves the diag-
nostic features of the species. As in the sample of upper molars from Powder Wash
(Krishtalka and Stucky, 1983Z?), the paracone is slightly lower than the metacone, but
higher than the protocone and stylar cusp B. Conules and stylar cusps other than B are
absent or were worn away.

— Setoguchi (1973) and Bown (1979) correctly allocated
Peratherium chested to Peradectes, based on the entoconid-hypocon-
ulid configuration on the lower molars. These authors and Rose (1981)
also considered P. chested a senior synomym of P. protinnominatus,
a conclusion that is at odds with our analysis of the material. As
discussed elsewhere (Krishtalka and Stucky, 1983a), P. protinnomi-
natus is a valid species and includes the Clarkforkian and early Wa-
satchian material from the Bighorn Basin that Bown (1979) and Rose
(1981) referred to P. chesteri.

The most complete known specimen of P. chesteri is UW 984, a
lower jaw with P3-M4 from Tabernacle Butte, which was previously
identified as Peratherium cf. innominatum (McGrew, 1959) and Per-
adectes sp. cf P. innominatus (Bown, 1982). P. innominatum was
shown above to be a species of Peratherium rather than Peradectes,
and UW 984 bears the hypoconulid-entoconid complex of the latter.
M3 on UW 984, as on the type of P. chesteri, is smaller than that of
P. elegans and has a narrower talonid than trigonid. It is also smaller
than M3 of P. pauli and P. protinnominatus, and the buccal contour
at the base of the crown between the trigonid and talonid is less emar-
ginate. In addition, the morphology of P3-M4 on UW 984 indicates
that, in contrast to P. elegans, P. pauli and P. protinnominatus: molars
of P. chesteri have lower trigonids; P3 is lower than Mi, Mi_3 are
narrower in proportion to length, have narrower talonids and a more
nearly straight buccal margin basally; and M4 has a shorter talonid than
trigonid. Unlike the condition in P. elegans. Mi is smaller than M2. P.
chesteri differs from P. californicus in having a talonid on P3 and from
P. minutus in retaining the gap between M4 and the ascending ramus.

Apart from the single upper molar referred here, upper molars of P.
chesteri have been recovered from Powder Wash (Krishtalka and Stucky,
1983Z?) and the Green River Basin (West, 1973, Figs. C, D; PM 15682,
PM 15866). They differ from upper molars of all North American
species of Peradectes except P. minutus in having more transverse
M-“^, vestigial conules and stylar cusps C and D, and a more com-

1983

Krishtalka and Stucky— Wind River Faunas, 3

221

pressed protocone, especially on M^. In contrast to P. minutus, in P.
chest eri have a higher paracone than protocone and stylar cusp

B, and a smaller stylar cusp B, and is less transverse, with a less
compressed protocone.

The features of the upper and lower dentition of P. chested, and its
temporal distribution, suggest that this species is intermediate mor-
phologically and phylogenetically between P. protinnominatus and P.
minutus.

ArmintodelphySy new genus

Etymology.— ^vmmXo, a hamlet in Natrona County, Wyoming; delphys, Gr., womb,
a common suffix for generic names of marsupials.

Type species.— Armintodelphys blacki, new specfes.

Included species. — TypQ species and A. dawsoni, new species.

Diagnosis. — Differs from all North American Tertiary didelphines
as follows: 1) entoconid lower and smaller than the hypoconulid, with
no entoconid notch; 2) talonid narrower (basally) than the trigonid on
Mi_2. Smaller than Mimoperadectes and with smaller paraconid than
metaconid on M2_4.

Known distdbution. — TdiXesX Wasatchian to earliest Bridgerian — Wind
River Basin (Wind River Fm.), Wyoming.

Discussion.— Armintodelphys is the hfth known genus of North
American Tertiary didelphines and, along with Peradectes and Mim-
operadectes, is included in the Tribe Peradectini Crochet, 1979 (as
emended here).

Armintodelphys is more derived than species of Peradectes in having
an entoconid that is much smaller and lower than the hypoconulid on
M2_4. Peradectes retains the primitive condition (for the Peradectini)
of an entoconid that is slightly taller than or subequal to the hypocon-
ulid. A revision of the species of Peradectes (Krishtalka and Stucky,
1983a) indicates that Armintodelphys most closely resembles Pera-
dectes pauli in the buccal emargination of the base of the crown between
the trigonid and talonid of M2_3, the longer talonid than trigonid on
M4, and the narrower talonid than trigonid on M3. These features imply
that Armintodelphys evolved from or shared a common ancestry with
P. pauli.

Armintodelphys blacki, new species
(Fig. 5; Table 1)

Etymology.— XgdLvaQd in honor of Craig C. Black for his contribution to the paleontology
of the Wind River Basin.

Type. — CM 41159, partial left dentary with M3_4 and alveoli for P3-

M^.

222

Annals of Carnegie Museum

VOL. 52

Type locality.— CM. loc. 1039 (Buck Spring: Lost Cabin K-5), Lost
Cabin Member, Wind River Formation, Natrona County, Wyoming.

Referred specimens. — DenXary fragment with M3 talonid— CM 41 161; associated P3,
Ml trigonid, M2, M3 trigonid— UCM 45252.

Localities.— TypQ locality (Lostcabinian, Lost Cabin Member) and
UCM loc. 80061 (Gardnerbuttean, Lost Cabin Member), both in the
Wind River Basin (Wind River Fm.), Wyoming.

Known distribution. — hate Wasatchian to earliest Bridgerian — Wind
River Basin (Wind River Fm.), Wyoming.

Diagnosis. - Largest species of Armintodelphys; no posterior cingulid
on M2_3.

Description. — Upper molars of A. blacki have not been recovered. The dentary, long
and slender as in other didelphines, is 4.5 mm deep below M4. The inflected angle begins
just below the coronoid process and the mental foramen occurs directly below the pos-
terior root of P3. Alveoli preserved on the type specimen indicate that there were no
diastemata between the anterior premolars. P3 is premolariform. The single, dominant
cusp— the protoconid— bears a posterolabial crest that extends from its apex to the labial
border of the postcingulid, as in Peradectes and Peratherium, but in contrast to Mim-
operadectes where this crest is posteromedial (Down and Rose, 1979, PI. 1, Fig. 2). No
talonid cusps are present, but a talonid basin is developed posterolingual to the protoconid
and anterior to the strong postcingulid.

The cusps on the trigonid fragment of M, are similar in size and position to M, of
Peratherium and Peradectes. The paraconid is anterior and almost as high as the meta-
conid, which is directly lingual to and lower than the protoconid. Compared to M,, the
trigonids on M2 and M3 are wider and more compressed anteroposteriorly. A groove on
the anterior face of the paraconid receives the hypoconulid of the anterior molar. On
M2^ the talonid is narrower than the trigonid and the cristid obliqua meets the posterior
wall of the trigonid labial to the protocristid notch. The buccal base of the crown is
excavated between the talonid and trigonid. A wear facet in the hypoflexid notch descends
the posterior face of the trigonid to the dento-enamel border of the tooth, suggesting
that, as in A. dawsoni, the paracone on the upper molars was as tall as the metacone.
The entoconid, smaller than in Peratherium, Peradectes, and Mimoperadectes, and well-
removed from the metaconid, occurs on a lingual ridge that runs from the metaconid
to the hypoconulid. The latter is the tallest talonid cusp and lies posterolabial to the
entoconid. On M2_3 the talonid notch is lower than the point at which the cristid obliqua
meets the trigonid. M4, known only in the type, resembles M2_3 but has a more elongate
talonid, with a more medial and posteriorly projecting hypoconulid.

Armintodelphys dawsoni, new species
(Fig. 5; Table 1)

Etymology . —U2imedL in honor of Mary R. Dawson for her contributions to the pa-
leontology of the Wind River Basin.

Type. — CM 55569, partial dentary with talonid of M2, M3, and tri-
gonid of M4.

Type locality. — CM loc. 34 (Davis Ranch, also called Sullivan Ranch;
“maroon shale” layer of Guthrie, 1971), Lost Cabin Member, Wind
River Formation, Wind River Basin, Natrona County, Wyoming.

1983

Krishtalka and Stucky—Wind River Faunas, 3

223

Fig. 5.—Armintodelphys, new genus. (A) A. dawsoni, new species, CM 55569, RM2_4
(type); (B) A. blacki, new species, CM 41159, LM3^ (type); both approx. X 15.

224

Annals of Carnegie Museum

VOL. 52

Referred specimen. — M2— CM 55560.

Localities.— TyipQ locality (Gardnerbuttean) only.

Known distribution. — Bridge rian — Wind River Basin (Wind
River Fm.), Wyoming; Uinta Basin (Green River Fm.), Utah.

— Smallest species of Armintodelphys; differs from A. blacki
in having a postcingulid on M2_3.

Description.— depth of the mandible below M4 is 2.7 mm in the type specimen.
Apart from the differences cited in the diagnosis, the lower molars compare favorably
to those of A. blacki described above.

Occurrences and Relationships

Three lineages of Peratherium—P. comstocki, P. knighti, and P. in-
nominatum— differing mainly in size, appear abruptly in the early Wa-
satchian; P. comstocki extends to the early Bridgerian, and the other
two into the Duchesnean. A fourth, P. marsupium, ranges from the
middle Wasatchian to the Duchesnean, and a fifth, P. edwardi, is known
only from the late Wasatchian. Anagenetic change in these species is
small or undetectable; four of the species are stable, whereas P. in-
nominatum appears to undergo a slight increase in size. Three of these
species (P. innominatum, P. marsupium, P. comstocki) are present in
the Wind River Formation. They occur in lithosympatry at only two
of the eleven localities that yielded marsupial material (CM loc. 34;
UCM loc. 80062).

The Peradectini are represented in the Wind River Formation by
two species in a new genus, Armintodelphys (A. blacki and A. dawsoni),
and by Peradectes chest eri. Armintodelphys is most closely related to
Peradectes pauli. P. chesteri, known from late Wasatchian to late Brid-
gerian horizons, appears to be morphologically and phylogenetically
intermediate between P. protinnominatus and P. minutus. If this in-
ference is correct, trends in the P. protinnominatus-P. chesteri-P. mi-
nutus lineage involved the development of more transverse with
a more compressed protocone, and reduction in the crown height of
the upper and lower molars, in the conules and stylar cusps C and D,
in the length and width of the talonid on the lower molars and in the
buccal emargination between the talonid and trigonid on Mi_3.

Acknowledgments

We thank Mary R. Dawson for discussions concerning the Wind River Formation
marsupials and for reviewing the manuscript, and Richard C. Fox for sharing an un-
published manuscript on Herpetotherium. The following kindly provided access to mar-
supial material in their care: Donald Baird (PU), Robert Emry (USNM), Howard Hutch-
ison (UCMP), Malcolm McKenna (AMNH), John Ostrom (YPM) and Peter Robinson
(UCM). We are grateful to Nick Piesco (University of Pittsburgh, School of Dentistry)
for use of the SEM facilities and his technical assistance. This work was supported in

1983

Krishtalka and Stucky— Wind River Faunas, 3

225

part by grants from the M. Graham Netting Research Fund of the Carnegie Museum of

Natural History through a grant from the Cordelia Scaife May Charitable Trust, the Rea

Postdoctoral Fellowship, Carnegie Museum of Natural History, and the Walker Van

Riper Fund of the University of Colorado Museum.

Literature Cited

Archibald, J. D. 1982. A study of Mammalia and geology across the Cretaceous-
Tertiary boundary in Garfield County, Montana. Univ. California Publ. Geol. Sci.,
122:U286.

Aymard, A. 1 846. Essai Monographique sur un genre nouveau de mammiferes fossile
trouve dans le Haute-Loire et nomme Entelodon. Ann. Soc. Agr. du Puy, 12:227-
267.

. 1850. Compte rendu de la seance du 13 avril 1849, reponse a M. Robert sur

les mammiferes fossiles des calcaires du Puy. Ann. Soc. Agr. du Puy, 14:80-86.

Down, T. M. 1979. Geology and mammalian paleontology of the Sand Creek facies.
Lower Willwood Formation (Lower Eocene), Washakie County, Wyoming. Mem.
Geol. Survey Wyoming, 2:1-151.

. 1982. Geology, paleontology, and correlation of Eocene volcaniclastic rocks,

southeast Absaroka Range, Hot Springs County, Wyoming. Geol. Survey Prof Pap.
1201-A:1-75.

Bown, T. M., AND K. D. Rose. 1979. a new marsupial, and Worlandia,

a new dermopteran, from the lower part of the Willwood Formation (Early Eocene),
Bighorn Basin, Wyoming. Univ. Michigan Contribs. Mus. Paleont., 25:89-104.

Clemens, W. A. 1979. Marsupialia. Pp. 192-220, in Mesozoic mammals, the first two-
thirds of mammalian history (J. A. Lillegraven, Z. Kielan-Jaworowska, and W. A.
Clemens, eds.), Univ. California Press, Berkeley, x + 31 1 pp.

Cope, E. D. 1873. Third notice of extinct Vertebrata from the Tertiary of the plains.
Paleont. Bull., 16:1-8.

. 1880. The northern Wasatch fauna. Amer. Nat., 14:908-909.

. 1884. The Vertebrata of the Tertiary formations of the West. Report U.S. Geol.

Surv. Terr., F. V. Hayden, Washington, D.C., 3:1-1009.

Crochet, J-Y. 1977. Les Didelphidae (Marsupicarnivora, Marsupialia) holarctiques
tertiares. C. R. Acad. Sci. Paris, 284, serie D:357-360.

. 1979. Diversite systematique des Didelphidae (Marsupialia) Europeens Ter-
tiares. Geobios, 12:365-378.

Delson, E. 1971. Fossil mammals of the Wasatchian Powder River local fauna. Eocene
of northeast Wyoming. Bull. Amer. Mus. Nat. Hist., 146:309-364.

Eaton, J. G. 1 982. Paleontology and correlation of Eocene volcanic rocks in the Carter
Mountain area. Park County, southeastern Absaroka Range, Wyoming. Univ. Wyo.
Contribs. Geol., 21:153-194.

Filhol, H. 1879. Etude du mammiferes fossiles de Saint-Gerand-le-Puy. Ann. Sc.
Geol. Paris, 10:1-253.

Gazin, C. L. 1952. The Lower Eocene Knight Formation of western Wyoming and its
mammalian faunas. Smithsonian Misc. Coll., 117:1-82.

. 1953. The Tillodontia: an early Tertiary order of mammals. Smithsonian Misc.

Coll., 121:1-110.

. 1956. Paleocene mammalian faunas of the Bison Basin in south-central Wy-
oming. Smithsonian Misc. Coll., 131:1-57.

. 1962. A further study of the Lower Eocene mammal faunas of southwestern

Wyoming. Smithsonian Misc. Coll., 144:1-98.

Gingerich, P. D. 1979. Phylogeny of middle Eocene Adapidae (Mammalia, Primates)
in North America: Smilodectes and Notharctus. J. Paleont., 53:153-163.

226

Annals of Carnegie Museum

VOL. 52

. 1980. History of early Cenozoic vertebrate paleontology in the Bighorn Basin.

Pp. 7-24, in Early Cenozoic paleontology and stratigraphy of the Bighorn Basin,
Wyoming (P. D. Gingerich, ed.), Univ. Michigan Pap. Paleont., 24:vi + 1-146.
Gray, J. E. 1821. On the natural arrangement of vertebrose animals. London Med.
Reposit., 15:296-310.

Green, M., and J. E. Martin. 1976. Peratherium (Marsupialia: Didelphidae) from
the Oligocene and Miocene of South Dakota. Pp. 155-168, in ATHLON. Essays on
paleontology in honour of Loris Shano Russell. Royal Ontario Mus., Life Sci. Misc.
Publ., Toronto.

Guthrie, D. A. 1971. The mammalian fauna of the Lost Cabin Member, Wind River
Formation (Lower Eocene) of Wyoming. Ann. Carnegie Mus., 43:47-1 13.

Hay, O. P. 1930. Second bibliography and catalogue of the fossil Vertebrata. Volume
11. Carnegie Inst. Washington Publ., 390:1-1074.

ICrishtalka, L., and R. K. Stucky. 1 9^3>a. Paleocene and Eocene marsupials of North
America. Ann. Carnegie Mus., 52:229-263.

— . 19836. Middle Eocene marsupials from northeastern Utah. Ann. Carnegie

Mus., in press.

Lillegraven, j. a. 1976. Didelphids (Marsupialia) and Uintasorex (?Primates) from
later Eocene sediments of San Diego County, California. Trans. San Diego Soc. Nat.
Hist., 18:85-112.

Lucas, S. G., R. M. Schoch, E. Manning, and C. Tsentas. 1981. The Eocene bio-
stratigraphy of New Mexico. Bull. Geol. Soc. Amer., pt. 1, 92:951-967.

Marsh, O. C. 1872. Preliminary description of new Tertiary mammals. Amer. J. Sci.,
ser. 3, 4:1-35.

Matthew, W. D. 1899. A provisional classification of the freshwater Tertiary of the
West. Bull. Amer. Mus. Nat. Hist., 12:19-75.

. 1909. Faunal lists of the Tertiary Mammalia of the West. Bull. U.S. Geol.

Surv., 361:91-138.

Matthew, W. D., and W. Granger. 1921. New genera of Paleocene mammals. Amer.
Mus. Novitates, 13:1-7.

McGrew, P. O. 1937. New marsupials from the Tertiary of Nebraska. J. Geol., 45:
448-455.

. 1959. Marsupialia. Pp. 147-148, in The geology and paleontology of the Elk

Mountain and Tabernacle Butte area, Wyoming. Bull. Amer. Mus. Nat. Hist., 1 17:
117-176.

McGrew, P. O., and R. Sullivan. 1970. The stratigraphy and paleontology of Bridger
A. Univ. Wyoming Contrib. Geol., 9:66-85.

McKenna, M. C. 1960. Fossil Mammalia from the early Wasatchian Four Mile local
fauna. Eocene of northwest Colorado. Univ. California Publ. Geol. Sci., 37:1-130.
Osborn, H. F. 1909. Cenozoic mammal horizons of western North America, with
faunal lists of the Tertiary mammals of the West by William Diller Matthew. Bull.
U.S. Geol. Surv., 361:1-138.

Robinson, P. 1 966. Fossil Mammalia of the Huerfano Formation, Eocene of Colorado.
Bull. Yale Peabody Mus. Nat. Hist., 21:1-95.

. 1968. Nyctitheriidae (Mammalia, Insectivora) from the Bridger Formation of

Wyoming. Univ. Wyoming Contrib. Geol., 7:129-138.

Rose, K. D. 1981. The Clarkforkian Land-Mammal Age and mammalian faunal com-
position across the Paleocene-Eocene boundary. Univ. Michigan Pap. Paleont., 26:
1-197.

ScHANKLER, D. M. 1980. Faunal zonation of the Willwood Formation in the central
Bighorn Basin, Wyoming. Pp. 99-1 14, in Early Cenozoic paleontology and stratig-
raphy of the Bighorn Basin, Wyoming (P. D. Gingerich, ed.), Univ. Michigan Pap.
Paleont., 24:vi + 1-146.

1983

Krishtalka and Stucky— Wind River Faunas, 3

227

Setoguchi, T. 1973. Late Eocene marsupials and insectivores from the Tepee Trail
Formation, Badwater, Wyoming. Unpublished Ph.D. dissert., Texas Tech Univ.,
Lubbock, 101 pp.

. 1975. Paleontology and geology of the Badwater Creek area, central Wyoming.

Part 1 1. Late Eocene marsupials. Ann. Carnegie Mus., 45:263-275.

Simpson, G. G. 1928. American Eocene didelphids. Amer. Mus. Novitates, 303:1-7.

. 1961. Principles of animal taxonomy. Columbia Univ. Press, New York,

247 pp.

. 1968. A didelphid (Marsupialia) from the early Eocene of Colorado. Postilla,

115:1-3.

Stock, C. 1936. Sespe Eocene didelphids. Proc. Nat. Acad. Sci., 22:122-124.

Stucky, R. K. 1982. Mammalian faunas and biostratigraphy of the upper part of the
Wind River Formation (early to middle Eocene), Natrona County, Wyoming, and
the Wasatchian-Bridgerian boundary. Unpublished Ph.D. dissert., Univ. Colorado,
Boulder, 278 pp.

Stucky, R. K., and L. Krishtalka. 1982. Revision of the Wind River faunas, early
Eocene of central Wyoming. Part 1. Introduction and Multituberculata. Ann. Car-
negie Mus., 51:39-56.

Troxell, E. L. 1923. A new marsupial. Amer. J. Sci., 5:507-5 10.

West, R. M. 1973. Geology and mammalian paleontology of the New Fork-Big Sandy
area, Sublette County, Wyoming. Fieldiana, 29:1-193.

. 1982. Fossil mammals from the Lower Buck Hill Group, Eocene of Trans-

Pecos Texas: Marsupicarnivora, Primates, Taeniodonta, Condylarthra, bunodont
Artiodactyla, and Dinocerata. Texas Mem. Mus. Pearce-Sellards Ser., 35:1-20.

West, R. M., and M. R. Dawson. 1973. Fossil mammals from the upper part of the
Cathedral Bluffs tongue of the Wasatch Formation (early Bridgerian), northern Green
River Basin, Wyoming. Univ. Wyoming Contrib. Geol., 12:33-41.

. 1975. Eocene fossil Mammalia from the Sand Wash Basin, northwestern Moffat

County, Colorado. Ann. Carnegie Mus., 45:231-253.

Back issues of many Annals of Carnegie Museum articles are
available, and a few early complete volumes and parts are listed
at half price. Orders and inquiries should be addressed to: Pub-
lications Secretary, Carnegie Museum, 4400 Forbes Avenue,
Pittsburgh, Pa. 15213.

73

ISSN 0097-4463

AN NALS

rfCARNECiE MUSEUM

CARNEGIE MUSEUM OF NATURAL HISTORY

4400 FORBES AVENUE • PITTSBURGH, PENNSYLVANIA 15213

VOLUMES! 16 SEPTEMBER 1983 ARTICLE 10

PALEOGENE AND EOCENE MARSUPIALS OF
NORTH AMERICA

Leonard Krishtalka
Associate Curator, Section of Vertebrate Fossils

Richard K. Stucky

Postdoctoral Fellow, Section of Vertebrate Fossils

Abstract

North American Paleocene and Eocene marsupials compose two tribes of didelphines.
The Didelphini contains five species of Peratherium {P. comstocki, P. edwardi, P. mar-
supium, P. knighti, P. innominatum). The Peradectini includes six species of Peradectes
{P. elegans, P. pauli, P. protinnominatus, P. californicus, P. chesteri, P. minutus), two
species of Armintodelphys (A. blacki, A. dawsoni), and one species of Mimoperadectes
{M. labrus). Previous allocations of A/phadon and Albertatherium to the Peradectini, P.
innominatum to Peradectes, P. protinnominatus to P. chesteri, and the latter to Herpe-
totheriurn are revised, as are the systematics of all of the species. The two tribes differ
in the structure of the entoconid-hypoconulid complex on the lower molars and in the
presence or absence of dilambdodonty on the upper molars. The five species of Pera-
therium show virtually no evolutionary change throughout their range in the Eocene. In
contrast, the six species of Peradectes form a branching lineage that is characterized by
cladogenetic speciation and anagenesis from the Paleocene through the Oligocene. Ar-
mintodelphys and Mimoperadectes appear to be early offshoots of that lineage.

Introduction

Marsupials occur in most early Tertiary North American faunas. The
first such record was Entomacodon minutus Marsh, 1872, from the
Bridger Basin, which was not identified as a marsupial until almost
one hundred years after its original description (Robinson, 1 968; Krish-
talka and Stucky, 1983<7). Similarly, Peratherium comstocki 1884,

Submitted 7 March 1983.

229

230

Annals of Carnegie Museum

VOL. 52

originally called a creodont, was first recognized as a didelphid by
Simpson (1928), who also reviewed the systematics of Peratherium
{j^Herpet other iuiri) marsupium (Troxell, 1923), and named Perather-
ium innorninatum from material that Matthew (\909b) had assigned
to this genus. Matthew (1899, 1909<2, 1909Z?) had debated the marsupial
versus insectivore affinities of E. minutus (including Centracodon de-
licatus Marsh, 1872) and P. comstocki and opted for the latter. This
position was colored by the late Nineteenth Century view, taken to its
extreme by Wortman (1901), of a elose relationship between marsu-
pials, ereodonts, and carnivores. Matthew (1909Z?:33 5-339) clarified
the distinctions between marsupials and allegedly closely related pla-
centals by identifying their “primitive” and “specialized” character-
istics; he (p. 321) emphasized the relative importance of derived fea-
tures in phylogenetic reconstructions. Indeed, Matthew’s discussion,
although devoid of cladistic terminology, presages what some current
students believe to be the modern innovation in systematics.

In 1921, Matthew and Granger described the first records of North
American Paleocene marsupials, Peradectes elegans and Thylacodon
pusillus. In the next 60 years Stock (1 936), McGrew (1 937, 1959), Gazin
(1952, 1956, 1962), McKenna (1960), Bown (1979), and Bown and
Rose (1979) added six species of Peratherium (P. californicum, P.
knighti, P. chesteri, P. edwardi, P. morrisi, P. macgrewi), two species
of Peradectes {P. pauli, P. protinnominatus), and two new genera, Nan-
odelphys and Mimoperadectes, each with one species {N. minutus, M.
labrus) to the North American Paleocene and Eocene didelphid record.
Most recently, we (Krishtalka and Stucky, 1983a) named a new genus,
Armintodelphys, with two species {A. blacki, A. dawsoni), from the
early and middle Eocene of Wyoming.

Partial systematic reviews of some of these 1 9 species in nine genera
have occurred only within the last ten years (Setoguchi, 1973, 1975;
Bown, 1979, 1982; Lillegraven, 1976; Crochet, 1977, 1979), resulting
in a confusing, and often conflicting, shuffling of species among Per-
atherium, Herpetotherium, Peradectes, and Nanodelphys. None of these
studies encompassed all Paleocene and Eocene North American di-
delphids. In the course of describing the marsupials from the Wind
River Formation (Krishtalka and Stucky, 1983a) it became apparent
that such a study was necessary. Those didelphids recorded from the
Wind River Formation {Peratherium comstocki, P. marsupium, P. in-
nominatum; Peradectes chesteri; Armintodelphys blacki, A. dawsoni)
were systematically revised. Remaining North American Paleocene and
Eocene didelphids are revised here, except for Thylacodon pusillus,
which is under study elsewhere (see Clemens, 1979; Archibald, 1982).
Also included is a summary of the occurrences of these didelphids and
a reconstruction of their evolutionary relationships.

1983

Krishtalka and Stucky — Paleogene and Eocene Marsupials

231

Abbreviations used in this paper are as follows; AMNH, American Museum of Natural
History; CM, Carnegie Museum of Natural History; KU, University of Kansas; LACM,
Los Angeles County Museum; MCZ, Museum of Comparative Zoology, Harvard Uni-
versity; PM, Field Museum of Natural History; PU, Princeton University (Museum);
TTU-P, Texas Tech University, Paleontology; UCM, University of Colorado Museum;
UCMP, University of California Museum of Paleontology; UM, University of Michigan
Museum of Paleontology: USNM, U.S. National Museum; UW, University of Wyoming;
YPM, Yale Peabody Museum; Fm., Formation; L, length; W, width.

Systematics

Family Didelphidae Gray, 1821
Subfamily Didelphinae (Gray, 1821)

Crochet (1979) separated known didelphines into two groups, the
Didelphini and Peradectini, based, in part, on diagnostic features of
the dentition that Setoguchi (1973) had identified earlier. These fea-
tures, as well as those identified in another study (Krishtalka and Stucky,
1983(2), indicate that members of the Didelphini have lower molars
with a tall, spire-like entoconid, a lower, proximal and posterior hy-
poconulid, and a deep entoconid notch. Their upper molars are di-
lambdodont, with a lower paracone than metacone, strong conules and
stylar cusps, and a posteriorly expanded protoconal base. In contrast,
members of the Peradectini have lower molars with subequal and
twinned entoconid and hypoconulid, and a weak or vestigial entoconid
notch. Their upper molars are not dilambdodont and bear subequal
paracone and metacone, a V-shaped protocone, and weak or vestigial
stylar cusps and conules.

Both groups differ from known Cretaceous didelphines in having
lower molars with shorter talonids and a cristid obliqua that meets the
posterior wall of the trigonid labial to the protocristid notch. This
implies a common ancestry for the Didelphini and Peradectini from a
Cretaceous didelphine. These features, as well as those that are diag-
nostic of the Peradectini, also imply that Alphadon and Albert at herium
do not belong in that tribe (contra Crochet, 1979).

Tribe Didelphini Crochet, 1979

According to Crochet (1979), Herpetotherium is the only North
American Tertiary genus in this tribe and includes all North American
species formerly referred to Peratherium. H. fugax, the type species,
has a diagnostic anterior dentition (Fox, manuscript; personal com-
munication, 1982). However, the characters of the molar dentition
cited by Crochet (1977) in support of this synonymy are variable, at
least within the North American Eocene species of Didelphini (Krish-
talka and Stucky, 1983(2) — species for which the anterior dentition is
not known. Accordingly, these species are retained in Peratherium.

232

Annals of Carnegie Museum

VOL. 52

Peratherium Aymard, 1846

This genus is not known from North American Paleocene horizons.
Based on the known material, the five North American Eocene species
of Peratherium are, in decreasing order of size (Tables 1 and 3), P.
comstocki Cope, 1884, P. edwardi Gazin, 1952, P. marsupium, (Trox-
ell, 1923), P. knighti McGrew, 1959, and P. innominatum Simpson,
1928.

The systematics of three of these species— P. comstocki, P. marsu-
pium, and P. innominatum— reviewed in detail in a treatment of
the marsupials from the Wind River Formation (Krishtalka and Stucky,
1983(2), and are briefly summarized here.

Peratherium comstocki Cope, 1884

Referred specimens. ~\n addition to material referred elsewhere (Cope, 1 884; Simpson,
1928, 1968; West, 1982; Krishtalka and Stucky, 1983a), CM 13901 (M3_4), from the
Bridger Fm., Twin Buttes, Bridger Basin, Wyoming.

Discussion.— P. comstocki is the largest known North American
Eocene species of Peratherium. The referral of CM 13901 and material
from Agua Fria, Texas (West, 1982) extends the known range of this
species from the early Wasatchian to the late Bridgerian and perhaps
into the earliest Uintan, during which time preserved dentitions of P.
comstocki show no significant morphological change.

Peratherium marsupium (Troxell, 1923)

Referred specimens.— In addition to material referred elsewhere [Troxell, 1 923; McGrew
and Sullivan, 1970; West, 1973 (except PM 15864); West and Dawson, 1975 (except
CM 23194); Setoguchi, 1975 (except CM 23793, 23803, TTU-P 1375, 1238); Eaton,
1982; West, 1982; Krishtalka and Stucky, 1983a. 1983Z?], CM 36533 (DPh, TTU-P
5966, CM 29061 (both M'), CM 15043, 15144, 15668, 15708, 29109, 51345, 53754,
51422 (all M^), CM 28850, 29021 (both Mh, CM 15626 (MT CM 55938 (P3-M,), CM
55939 (P3), CM 29045, 55940, 55941 (M^ or M3) from localities 5A, 5 Front, 6, Rodent,
Wood and 20, Wagon Bed Fm., Wind River Basin, Wyoming.

Discussion. — This, species is well-known from middle Wasatchian to
Duchesnean faunas of the western interior of North America, and, like
P. comstocki, undergoes no perceptible changes in preserved parts of
the dentition during its geological extent.

Peratherium innominatum Simpson, 1928

Referred specimens. — In addition to material referred elsewhere [Simpson, 1928; Bown,
1979 (as P. macgrewi, except UW 10129); Eaton, 1982; Krishtalka and Stucky, 1983a,
1983Z)], UCMP 44095 (M'-T CM 42137 (M2 or M3), from Sand Quarry, Four Mile area,
Wasatch Fm., Colorado; part of UW 9742 (M9, from the Willwood Fm., Bighorn Basin,
Wyoming; UCMP 109647 (Mh, from the Uintan of California; CM 23851, 23852 (both
DPT CM 15728 (M*), CM 15083, 23787, 23918, 51354 (all MT CM 23788, 51237
(both M2-T TTU-P 1349, CM 14644, 15034, 15035, 15049, (all MT CM 23832, 23893

Table Dimensions of type specimens of species o/Peratherium.

1983

Krishtalka and Stucky — Paleogene and Eocene Marsupials

233

234

Annals of Carnegie Museum

VOL. 52

(both M^), TTU-P 1360, CM 15676, 15682, 16004, 23935, 23936 (all DP3), TTU-P
2420, CM 15121, 15132, 15692 (all M,), TTU-P 2425, 2429, CM 15078, 15102, 15672,
15729, 23844, 23887, 23901, 23906, 23917, 51248 (all M2 or M3), CM 23846 (MJ,
from localities 5, 5A, 5 Front, 5 Back, 6, Wood and 20, Wagon Bed Fm., Wind River
Basin, Wyoming.

Discussion. P. comstocki and P. marsupium, P. innomi-
natum has been confused with other didelphines. This species was
considered a species of Peradectes by Setoguchi (1973), Bown (1982),
and Crochet (1979), due, perhaps, to a misinterpretation of the diag-
nostic (and damaged) features on the type specimen. M2 on the latter
(AMNH 11493) preserves a large, spire-like entoconid, a lower, pos-
terior hypoconulid, and a deep entoconid notch. These features are
also evident on the broken M3 on the type and are diagnostic of Per-
atherium and other Didelphini. P. innominatum is closest in size to
contemporaneous Peradectes chesteri, the type of which has an M3 with
subequal entoconid and hypoconulid and a weak entoconid notch.
Upper molars of these two species from the same localities also show
the diagnostic features of Peratherium and Peradectes. Unlike the latter,
upper molars of P. innominatum are dilambdodont, with a much lower
paracone than metacone, prominent stylar cusps and conules, and a
posterior expansion of the base of the protocone.

Study of the sample of P. innominatum from the Wind River For-
mation (Krishtalka and Stucky, 1983dz) and Powder Wash (Krishtalka
and Stucky, 1983Z?) indicates that this species includes the Graybullian
P. macgrewi Bown, 1979, as a temporal subspecies. Two of Bown’s
identifications are revised: one of the upper molars in UW 9742 (Bown,
1 979, Fig. 40a, center) is dilambdodont and belongs in P. innominatum
(rather than P. chesteri)-, UW 10129 (Fig. 40c) is not dilambdodont
and is transferred to Peradectes protinnorninatus (from P. macgrewi).

Remains of this species have also been identified (Krishtalka and
Stucky, 1983^2) in the early Eocene Four Mile, middle Eocene Carter
Mountain (Eaton, 1982) and late Eocene Badwater samples of didel-
phids (part of Setoguchi’s, 1975, Peratherium cf. P. knighti and Nan-
odelphys cf. N. minutus). Lillegraven’s (1976) referred material of Per-
atherium sp. cf P. knighti from the Uintan of San Diego County may
include specimens of both P. knighti and P. innominatum. The range
in size of the molars overlaps with that of both species, and the figured
upper molars include two morphs: some have a shallow ectoflexus
(Lillegraven, 1976, Fig. 1, 2), whereas one (Fig. 3c) has a deep ecto-
flexus, as is characteristic of P. knighti and P. innominatum, respec-
tively.

Bown (1982) referred material from the Aycross Formation to Per-
adectes sp. cf P. innominatus on the basis of their similarity to UW
984 from Tabernacle Butte (see McGrew, 1959). UW 984 is identified

1983

Krishtalka and Stucky— Paleogene and Eocene Marsupials

235

here and elsewhere (Krishtalka and Stucky, 1983(2) as Peradectes ches-
teri, according to features discussed below; the Aycross material, by
implication, may also belong to P. chested.

According to these observations, the known record of P. innomi-
natum extends from the early Wasatchian to the Duchesnean of the
western interior of North America, and possibly includes the Uintan
of California.

Pemtherium edwardi Gazin, 1952
(Fig. 1)

— USNM 19200, partial left dentary with M3_4.

Type locality. — 12 miles north of Big Piney, in SWl/4 section 33,
T. 32 N, R. Ill W, Sublette County, Wyoming.

Referred — USNM 19206 from the type locality.

Known distnbution. — TaXQ Wasatchian— Green River Basin (Wa-
satch Fm.), Wyoming.

Discussion. — This poorly known species is intermediate in size be-
tween P. comstocki and P. marsupium. Allocation of this material to
the former, as advocated by Setoguchi (1973) and Bown and Rose
(1979), is not warranted until a larger sample is available.

Pemtherium knighti McGrew, 1959
(Fig. 2)

Entomacodon minutus Marsh, 1872.

Centracodon delicatus Marsh, 1872.

Peratherium knighti McGrew, 1959.

Peratherium morrisi Gazin, 1962.

— AMNH 55684, partial right maxilla with M'“^.

Type locality. — Locality 5, Tabernacle Butte, upper part of Bridger
Formation, Wyoming.

Referred specimens. — In addition to material referred elsewhere [McGrew, 1959; West
and Dawson, 1973; Bown, 1982; Setoguchi, 1975 (except TTU-P 1237, 1349, 2425,
2429, CM 23787, 23832, 23935, 23936 -referred above to P. innominatum- TTU-P
5966 — referred above to P. marsupium)], YPM 13514, partial dentary with M4 (type of
Entomacodon minutus), YPM 13508, dentary with Pj.jMi^ (type of Centracodon de-
licatus), both from Henry’s Fork, Bridger Formation, Bridger Basin, Wyoming; UW 3003
(M4), UW 2988 (Mj^), from the lower part of the Bridger Fm., Bridger Basin, Wyoming;
PU 16115, partial dentary with M2_3 (type of Peratherium morrisi), from the Cathedral
Bluffs tongue of the Wasatch Fm., Washakie Basin, Wyoming; CM 42135 (Ma^), from
the type locality; CM 42139 (M,), UCMP 59131 (M2 or M3), from the Wasatch Fm.,
Four Mile area, Colorado; PM 15864 (M^^), from the lower part of the Bridger Fm.,
Green River Basin, Wyoming; CM 23194 (M2 or M3), from the Washakie Fm., Sand
Wash Basin, Colorado; CM 15600, 15601, 15606, 15656 (all DPT CM 15030, 15031,
15036, 15037, 15046, 15077, 15081, 15108, 15631, 15670, 23841 (all MO, CM 55936
(M‘'2-T CM 55937 (M^T CM 15048, 15029, 15097, 15105, 15608, 15614, 15634,

236

Annals of Carnegie Museum

VOL. 52

Fig. \.—Peratherium edwardi, USNM 19200, LM3_4, type; approx. X 14.

23803, 23888, 29069 (all M^), CM 15027, 15032, 15050, 15059, 15101, 15118, 29110
(all M3), CM 15033, 15079, 15080, 15143, 15636, 15678, 55963 (all M^), CM 15134
(DP3), CM 14642, 14645, 15028, 15058, 15114, 15119, 15150, 15741,51202 (all M,);
CM 15040, 15042, 15044, 15057, 15659, 15689, 15710, 15720, 15742, 23894, 23911,
28832, 28893, 29024, 55942, 55964, 55969 (all M^ or M3), from localities 5, 5A, 5
Front, 5 Back, 6, Wood and 20, Wagon Bed Fm., Wind River Basin, Wyoming.

Known distribution. — Early Wasatchian— Four Mile area (Wasatch
Fm.), Colorado. Bridgerian— Washakie and Green River basins (Ca-
thedral Bluffs tongue of the Wasatch Fm.), Bridger Basin (Bridger Fm.),
Green River Basin (upper and lower parts of the Bridger Fm.), Bighorn
Basin (Aycross Fm.), Wyoming. Late Bridgerian or Uintan— Sand Wash
Basin (Washakie Fm.), Colorado. Uintan and Duchesnean— Wind Riv-
er Basin (Wagon Bed Fm.), Wyoming.

Emended diagnosis. — Smaller than P. comstocki, P. edwardi, P.
marsupium; larger than P. innominatum; ectoflexus on much

shallower than in P. marsupium and P. innominatum.

Discussion. — Three taxa are conspecific with P. knighti-Entorna-
codon minutus, Centracodon delicatus, and Peratherium morrisi. The
first two were synonymized by Matthew {1909b) as E. minutus (which

1983

Krishtalka and Stucky— Paleogene and Eocene Marsupials 237

O

Fig. l.—Peratherium knighti. (A) YPM 13514, M4 (type of Entomacodon minutus)\ (B) CM 42135, RM3^ (from Tabernacle Butte, the
type locality); (C) PU 16115, RM2_3 (type of P. morrisi)\ all approx. X 15.

238

Annals of Carnegie Museum

VOL. 52

has page priority; Marsh, 1872:214-215), and the types of both were
tentatively identified as marsupials by Robinson (1968). The types of
E. minutus and P. morrisi are indistinguishable from lower dentitions
in the hypodigm of P. knighti from Tabernacle Butte. Peratherium has
priority over Entomacodon, as does E. minutus over P. knighti, but
the resultant P. minutum is preoccupied by P. minutum Aymard, 1 850.
C. delicatus is a nomen oblitum according to the rules of zoological
nomenclature (Art. 23b).

P. knighti is closest in size to the larger P. marsupium and the smaller
P. innominatum, but known samples of these species do not overlap
in size. Additionally, of P. knighti have shallower ectoflexi. Ac-
cording to these criteria, we have identified P. knighti from material
(see “Referred specimens”) previously assigned to: P. marsupium from
the lower part of the Bridger Formation, Green River Basin, Wyoming
(West, 1973) and the Washakie Formation, Sand Wash Basin (West
and Dawson, 1975); Peratherium sp. from Bridger A (McGrew and
Sullivan, 1970); as well as from unpublished specimens from the Four
Mile fauna, Colorado.

As discussed earlier, P. knighti may occur in the Uintan of California
(Lillegraven, 1976). Analysis of late Eocene samples of Peratherium
from Badwater indicates that material previously referred to Pera-
therium cf. P. knighti (Setoguchi, 1975) includes specimens of P. knigh-
ti, P. innominatum, and P. marsupium.

Based on these synonymies and reidentifications, P. knighti is known
from the early Wasatchian to the Duchesnean in the western interior
of North America, and possibly from the Uintan of California.

Tribe Peradectini Crochet, 1979

North American didelphines included here are Peradectes, Mimo-
peradectes, and Armintodelphys. Nanodelphys is congeneric with Per-
adectes (Crochet, 1978). As discussed under Didelphinae, and Evolu-
tionary Relationships, Alphadon and Albert atherium lack the diagnostic
characters of this tribe and (contra Crochet, 1979) are removed from
the Peradectini.

The systematics of one of the species of Peradectes {P. chesteri) and
both species of Armintodelphys (A. blacki and A. dawsoni) were treated
elsewhere (Krishtalka and Stucky, 1983^^), as part of a description of
the didelphines from the Wind River Formation. The systematics of
these taxa are briefly summarized here, and additional material is re-
ferred to P. chesteri.

Peradectes Matthew and Granger, 1921

Three North American species were originally assigned to Pera-
dectes—Xht type species, P. elegans Matthew and Granger, 1921; P.

1983

Krishtalkla and Stucky— Paleogene and Eocene Marsupials

239

pauli Gazin, 1956; and P. protinnominatus McKenna, 1960. Subse-
quently, Setoguchi (1973) and Bown (1979) identified Peratherium
chesteri Gazin, 1952, and Peratherium innominatum Simpson, 1928
as species of Peradectes, and P. protinnominatus as a junior synonym
of P. chesteri. Setoguchi (1973) also advocated inclusion of Peratherium
morrisi in Peradectes, and Bown (1979) noted a lack of sufficient mor-
phological distance between Nanodelphys minutus, the type species,
and Peradectes chesteri to warrant generic distinction. Lillegraven (1 976)
transferred Peratherium californicum Stock, 1936, to Nanodelphys, and
opted for retaining N. californicus and N. minutus despite an apparent
lack of morphological distinction. Crochet (1 978) reduced Nanodelphys
to a subgenus of Peradectes in which he included the North American
species P. minutus, P. californicus, P. innominatus, and P. protinnom-
inatus. P. elegans was referred to the subgenus Peradectes, P. pauli was
considered too poorly known to be discussed, and P. chesteri was
assigned to Herpetotherium. Lastly, Clemens (1979) and Archibald
(1982) have suggested that Thylacodon pusillus is a species of Pera-
dectes.

Review of the type and referred material of these taxa indicates that
Peradectes is characterized by M,_3 with short talonids, labial cristid
obliquas, and an entoconid and hypoconulid that are low, subequal,
and closely appressed; these cusps are separated by a weak entoconid
notch and share a common internal talonid wall. On the paracone
and metacone are subequal and not dilambdodont, the stylar cusps
and conules are weak, and the posterolingual part of the base of the
protocone is not expanded. These features are also diagnostic of the
Peradectini. Compared to Cretaceous didelphines, the absence of di-
lambdodonty, and the V-shaped protocone are primitive retentions.
Unlike Mimoperadectes, the metaconid is larger and higher than the
paraconid on Mi_3 of Peradectes^ unlike Armintodelphys, the entoconid
is subequal to the hypoconulid, rather than reduced to a nubbin. Mea-
surements of these taxa are given in Tables 2 and 3.

Systematic conclusions implied by examination of the material are:
(1) Nanodelphys is congeneric with Peradectes] as suggested by Bown
(1979) and Crochet (1978), the morphological distinctions between N.
minutus, the type species, and species of Peradectes are of specific
magnitude and are minor eompared to those that define other genera
of didelphines; Crochet’s (1978) subgenus Nanodelphys has no system-
atic utility; (2) there are six discernable North American species of
Peradectes: P. elegans, P. pauli, P. protinnominatus, P. chesteri {=Pera-
therium chesteri), P. californicus {=Peratherium californicum and Nan-
odelphys californicus) and P. minutus {=Nanodelphys minutus)] (3) nei-
ther Peratherium morrisi nor P. innominatum belongs in Peradectes]
the former is a junior synonym of Peratherium knight i and the latter

Table 1. — Dimensions of type and paratype specimens (lower dentition) of species o/Peradectes, Armintodelphys, and Mimoperadectes.

240

Annals of Carnegie Museum

VOL. 52

o o

Cl.

d d

OJ

a

d

"a2 r3

D. b

(U cu <u

a a a

>>>>>>

r- oo

OO 00

o o

^ Tt m

0^ On ON
m IT)

H

5

fN (N

^

K

Oh Cu Ph

:z

2: 2i

u

C/5 C/D

u u u

<

<

d d

d d d

hJ

<q

o

o

OO

On

oo

' '

' '

' '

d

d

d

o

IT)

'Nj-

<N

>rN

(N

(N

d

o

ir>

o

m

o

o

ON

O

o

oo

oo

oo

d

d

d

o

o

o

r-'

-J

Co

NO

NO

NO

o

s:

S

■S

d

5

o

s

d

-Si

G

s

c$

sc.

o

o

s;

•S

o

o

i'

NO

No

-C)

%

!h

~C)

NO

Q-;

1.0

1.0

oo

d

1

ON

d

0.9

0.8

No

d

a,’

d

1983

Krishtalka and Stucky— Paleogene and Eocene Marsupials

241

Table 3. —Dimensions of type and paratype specimens (upper dentition) of species of
Peradectes, Mimoperadectes, and Peratherium.

M'

M

3

M

Taxa and catalog no.

L W

L W

L

w

L

w

AMNH 17369 (paratype)

Peradectes elegans

1.60 1.50 1.60 1.70

1.40

1.90

1.05

1.95

UCMP 44077

Peradectes protinnominatus

1.6 1.4 1.5 1.7

1.4

2.0

UM 66144

Mimoperadectes labrus
2.84 3.09 3.47

2.94

3.66

2.36

3.26

AMNH 55684

Peratherium knight i

1.9 1.6 2.05 2.0

2.05

2.2

is a valid species of Peratherium (see above); (4) Thylacodon pusillus
may not belong in Peradectes or the Peradectini.

Peradectes elegans Matthew and Granger, 1921
(Fig. 3)

Typp.— AMNH 17376, paired dentaries with RP,, P3-M4 and
LP2-M4.

Paratype.— AMNH 17369

Type locality. — MsLSon Pocket, Animas Formation (see Lucas and
Ingersoll, 1981), Colorado.

Referred specimens.— In addition to the material referred elsewhere (Simpson, 1935;
Gazin, 1 956), UCMP 44767 (M2^), CM 42 1 38 (M3), from the Wasatch Fm., Sand Quarry,
Four Mile area, Colorado; CM 41108 (M2_3), UCM 46216 (M3_4), from the Fort Union
Fm., Saddle locality. Bison Basin, Wyoming: AMNH 17199 (M'-^), from the type locality.

Known distribution. — — Basin (Fort Union Fm.), Wy-

oming; San Juan Basin (Animas Fm.), Colorado. Early Wasatchian —
Four Mile area (Wasatch Fm.), Colorado.

Emended diagnosis. — from other species of Peradectes as
follows; larger size; P3 more trenchant with longer talonid; P3 proto-
conid higher than that of molars; Mj as large as M^; Mj_4 talonids
broader and longer; M3 talonid broader than trigonid; M*"^ with strong-
er stylar cusps and conules and with uncompressed protocone.

Discussion.— ThQ oldest known material of P. elegans, from Bison
Basin, includes M2_4 that are broad in proportion to length, with wider
talonids than trigonids. Lower molars on the type of P. elegans from
the Tiffanian Mason Pocket are similar; curiously, RM3 on the type
has a slightly narrower talonid than LM3. Mi on the type is as large as

242

Annals of Carnegie Museum

VOL. 52

1983

Krishtalka and Stucky— Paleogene and Eocene Marsupials

243

Fig. 3. — Peradectes elegans. (C) AMNH 17369, paratype; approx. X 16.5.

M2. P3 is trenchant, higher than the molars and has a long talonid. On
M3 the talonid is broader than the trigonid. These characters distinguish
the lower dentition of P. elegans from that of all other North American
species of Peradectes and indicate that UCMP 44767, formerly iden-
tified as P. protinnorninatus (McKenna, 1960, Fig. 1 7c), and CM 42138,
both from Four Mile, represent P. elegans. M3 in these specimens has
a wider talonid than trigonid; M2_4 in UCMP 44767 are indistinguish-
able from P. elegans in size and structure, and the distance between
the roots of Mj implies that M, was as large as, if not larger than, M2.
This material represents the first Wasatchian record of P. elegans.

Upper molars of P. elegans from Mason Pocket are not transverse,
but in occlusal view resemble an equilateral triangle. As noted by
Simpson (1935), stylar cusps B, C, and D are present although variably
developed, the conules are moderately strong and the paracone, slightly
lower than the metacone, is higher than the protocone. In contrast, all

244

Annals of Carnegie Museum

VOL. 52

Other North American species of Peradectes have upper molars that
are more transverse, with an anteroposteriorly compressed protocone
and weaker or vestigial conules and stylar cusps C and D. Upper molars
of P. elegans are not known from Four Mile.

Peradectes pauli Gazin, 1956
(Fig. 4)

— USNM 20879, partial left dentary with M3_4.

Paratvpe. — \JSNM 20880, partial left dentary with M, and part
of M..

Type locality. — Saddle locality. Bison Basin, Fort Union Formation,
Wyoming.

Referred specimens. — In addition to material referred elsewhere (Gazin, 1956), CM
41109 (MO, MCZ 20793 (M^'O, from the type locality.

Known distribution. — Ti^aman—Bi^on Basin (Fort Union Fm.), Wy-
oming.

Emended diagnosis. — Compared to P. elegans: smaller size; Mi
smaller than M^; M3 with narrower talonid than trigonid; M4 with
narrower talonid; M^“^ more transverse with compressed protocone.
Compared to P. protinnominatus, P. chesteri, P. californicus, and P.
minutus: bases of M3_4 more emarginate buccally between trigonid and
talonid; M4 talonid longer than trigonid. Compared to P. californicus,
P. chesteri, and P. minutus: wider talonid occlusally on Mi„3. Compared
to P. chesteri and P. minutus: Mi_3 wider in proportion to length; less
disparity in L/W ratio from M“ to M^; less compressed protocone on
M“"^; stronger conules and stylar cusps C and D. Compared to P.
minutus: paracone proportionately higher than protocone and stylar
cusp B.

— Curiously, P. pauli has been ignored in recent impor-
tant studies of early Tertiary didelphines (McKenna, 1960; Setoguchi,
1973; Bown, 1979; Crochet, 1978; Rose, 1981).

In contrast to P. elegans, on the type and paratype of P. pauli Mi is
smaller than M2, M3_4 have narrower talonids, and the talonid on M3
is narrower than the trigonid. As such, P. pauli is a distinct species
from Bison Basin, where it occurs with P. elegans at the Saddle locality.

Analysis of the upper dentition leads to a similar conclusion. CM
41109, an isolated M^, and MCZ 20793, M“-^, both from Bison Basin,
resemble one another and differ from P. elegans in being more trans-
verse, with a narrower, anteroposteriorly compressed protocone and a
deeper ectoflexus. Except for slightly larger size, they are closely similar
to upper molars of P. protinnominatus from Four Mile and the Will-
wood Formation (see below; McKenna, 1960; Bown, 1979).

1983

Krishtalka and Stucky— Paleogene and Eocene Marsupials

245

Fig. A.—Peradectes pauli. (A) USNM 20880, LM, and trigonid of M2, paratype; (B)
USNM 20879, LM3_4, type; both approx. X 16.5.

246

Annals of Carnegie Museum

VOL. 52

P. pauIi is suitable morphologically and temporally to be basal to
the branching lineage that includes P. protinnominatus, P. californicus,
P. chesteri, and P. mi nut us.

Pemdectes protinnominatus McKenna, 1960

Peradectes chesteri BoyNn, 1979.

Peradectes cf. chesteri Rose, 1981.

Type. — VCMF 44077, partial right maxilla with

Paratvpes. -VCMP 45947 (M,_2), UCMP 45948 (M,_2), UCMP
45950 (P3-M1).

Type locality.— Alheit Pocket, Wasatch Fm., Four Mile area, Colo-
rado.

Referred specimens. — \n addition to material referred elsewhere (McKenna, 1960;
Down, 1979; Rose, 1981), CM 42138 (MO, UCMP 59132 (M3), UCMP 59130 (MO,
from the Wasatch Fm., Four Mile area, Colorado; UW 10129 (MO, from the Willwood
Fm., Bighorn Basin, Wyoming.

Known distribution. — Clderkforkidin to early Wasatchian — Bighorn
Basin (Willwood Fm.), Wyoming. Early Wasatchian — Four Mile area
(Wasatch Fm.), Colorado.

Emended diagnosis. — to P. elegans: smaller size; P3 tal-
onid shorter and trigonid lower than M^ Mj smaller than M2; M^~^
more transverse, protocone compressed. Compared to P. elegans and
P. pauli: M4 talonid shorter than trigonid; M3_4 with weaker emargi-
nation between trigonid and talonid; M-"^ protocone more anterior.
Compared to P. californicus, P. chesteri, and P. mi nut us: Mi_3 with
wider talonid occlusal width; M*“^ conules and stylar cusps C and D
stronger. Compared to P. chesteri and P. minutus: less disparity in
L/W ratio from M^ to M^; protocone on M^ less compressed; paracone
and metacone on M^“^ farther apart; Mi_3 wider in proportion to length.
Compared to P. minutus: M^“^ protocone less compressed; paracone
higher than protocone on M‘“^ and relatively higher than stylar
cusp B.

Z)/5CW55'Z6>«. — Setoguchi (1973) and Bown (1979) synonymized this
species with P. chesteri. Flowever, the type of the latter, an M3, can
only be compared to UCMP 44767, the only specimen with an M3 in
the original material of P. protinnominatus from Four Mile (McKenna,
1 960). UCMP 44767 is much larger than the type and referred material
of P. chesteri; it has a wider talonid than trigonid on M3, and subequal
Ml and M2, and was referred above to P. elegans.

In contrast to P. protinnominatus, the type and other specimens of
P. chesteri (see below) have lower molars that lack a buccal emargi-
nation between trigonid and talonid, and talonids that have a narrower

1983

Krishtalka and Stucky— Paleogene and Eocene Marsupials

247

occlusal width. As such, UW 9605 and UM 71663, from the Clark-
forkian and early Wasatchian, respectively, of the Bighorn Basin, more
closely resemble the paratypes of P. protinnominatus. Bown (1979) and
Rose (1981) reached the same conclusion concerning these specimens
(but allied them with P. chested = P. protinnominatus).

Upper molars of P. protinnominatus are also distinct from those of
P. chesteri. Diagnostic differences involve the degree of compression
of the protocone on M^, the disparity in L/W ratio from M' to M^,
and the development of the conules and stylar cusps C and D. As also
noted by Bown (1979), upper molars from his Bighorn Basin sample
are indistinguishable from the type of P. protinnominatus.

Upper molars in UCMP 44095 from Four Mile (McKenna, 1960,
Fig. 18a) and one of the upper molars in UW 9742 (Bown, 1979, Fig.
40a, center) are dilambdodont, and are reidentified here and elsewhere
(Krishtalka and Stucky, 1983^2) as Peratherium innominatum. UW
10129, originally identified as Peratherium macgrewi (Bown, 1979, Fig.
40c), is not dilambdodont and belongs in P. protinnominatus.

P. protinnominatus is intermediate in known dental morphology and
temporal occurrence between P. pauli and P. chesteri. It is also closely
related to P. californicus, which is more derived in the narrower occlusal
width of the molar talonids, and in the weaker stylar cusps C and D
and conules on M^“^.

Peradectes californicus (Stock, 1936)

Peratherium californicum Stock, 1936.

Nanodelphys cf. N. minutus Setoguchi, 1975, in part.

Nanodelphys californicus 1936), Lillegraven, 1976.

Peradectes californicum (Stock, 1936), Crochet, 1978.

Type. — (CIT) 202-1943, partial right dentary with P3-M2.

Type locality. — T ACM (CIT) loc. 202, Sespe Fm., Ventura County,
California.

Referred specimens. — In addition to material referred elsewhere (Lillegraven, 1976),
CM 15664 (Ml), CM 52421, 23845 (both M3), CM 52417 (MJ, CM 15138, 15612,
51367, 51250, 15611 (all M‘), CM 19748 (M^-^), CM 15671, 15104, 15149 (all MO, CM
15681 (MO, from localities 5, 5 Front, 5 Back, 6, 20, Wagon Bed Fm., Wind River Basin,
Wyoming.

Known distribution. — CinlMiSdin Diego County (Friars, Mission
Valley, ?Santiago, Sespe Fms.), California. Uintan to Duchesnean —
Wind River Basin (Wagon Bed Fm.), Wyoming.

Emended diagnosis. — CompdirQd to P. elegans: teeth much smaller;
P3 smaller than Ml M, smaller than M2; narrower talonid than trigonid
on M3; M^’"^ more transverse. Compared to P. elegans and P. paulf.
protocone more anterior on M‘“^; shorter talonid than trigonid on M4;

248

Annals of Carnegie Museum

VOL. 52

trigonids lower on Mi_4. Compared to P. elegans, P. pauli, and P.
protinnominatus: P3 without talonid (not known in P. pauli); Mi_3
talonids with narrower occlusal width; with weaker conules and
stylar cusps C and D. Compared to P. chesteri and P. rninutus: and

respectively, less transverse, with a less compressed protocone;
with a shallower ectoflexus and stronger stylar cusps C and D;
less disparity in L/W ratio from

Discussion. — lAWQgYZWQn (1976) recognized that Stock’s (1936) type
and referred material of Peratherium californicum represented an Uin-
tan species of Nanodelphys that was morphologically indistinguishable
from Nanodelphys cf. N. rninutus from the late Eocene of Badwater
(Setoguchi, 1973). He retained TV. californicus and N. rninutus on the
basis of geographic and temporal disparity. The two species, as under-
stood here, differ morphologically.

Setoguchi (1973) originally recognized two species of Nanodelphys
from Badwater, Nanodelphys cf. TV. rninutus and Nanodelphys sp. nov.,
which he later (Setoguchi, 1975) combined in the former. Crochet
(1978) and Bown (1979) independently suggested that Nanodelphys
and Peradectes were congeneric, and Crochet allocated TV. rninutus and
N. californicus to Peradectes.

From our analysis of the Badwater and Oligocene material it appears
that N. rninutus and TV. californicus are discernable species of Pera-
dectes. P. californicus includes some of the material that Setoguchi
(1973, 1975) assigned to Nanodelphys cf. TV. rninutus, whereas Seto- !
guchi’s (1973) Nanodelphys sp. nov. is referred below to Peradectes sp.
cf. P. rninutus. Of Setoguchi’s (1975) hgured specimens of Nanodelphys
cf. N. rninutus, TTU-P 2426 (Fig. 9) is a DP^ of Peratherium innorn-
inaturn and CM 16007 (Fig. 12) is an M“ of Peradectes sp. cf. P.
rninutus. The remainder are P. californicus. Some of his referred spec-
imens (TTU-P 1360, 2420, CM 23846) also represent Peratheriurr^
innorninaturn.

In summary, P. californicus is known from the Uintan of California,
and from the Uintan and Duchesnean of Wyoming where it occurs in '
lithosympatry with the smaller Peradectes sp. cf. P. rninutus at hve of
the Badwater late Eocene localities. The P. californicus material from
Wyoming shows a more restricted range in size than that from Cali-
fornia, possibly because of the co-occurrence of the two species at
Badwater and the implied character displacement. A larger sample of '
both species is needed to test this hypothesis.

P. californicus appears to have shared a common ancestry with the
P. chesteri-P. rninutus lineage from a P. protinnorninatus-\ikQ pera-
dectinine. The hrst three species are derived in having weaker conules
and stylar cusps C and D on the upper molars (see Setoguchi, 1975,

Fig. 11, 12; Lillegraven, 1976), and talonids on the lower molars with

1983

Krishtalka and Stucky— Paleogene and Eocene Marsupials

249

narrower occlusal widths between the cristid obliqua and the lingual
border. The cristid obliqua runs posteriorly from the trigonid before
turning labially to a more internal hypoconid, so that the occlusal width
of the talonid is approximately one-half of the basal width. In contrast,
other species of Peradectes retain the primitive posterolabial orienta-
tion of the cristid obliqua, a more buccal hypoconid, and an occlusally
wide talonid on the lower molars, as well as stronger conules and stylar
cusps C and D on the upper molars. in P. chested and in P.
rninutus (and Peradectes sp. cf P. minutus) are more advanced than
in P. californicus in having a more anteroposteriorly compressed pro-
tocone and vestigial or absent stylar cusps C and D. Also, the disparity
in L/W ratio from M* to is more marked.

Peradectes chesteri Gazin, 1952
(Fig. 5)

Referred specimens. — In addition to material referred elsewhere (Gazin, 1952; Krish-
talka and Stucky, 1983^2, 1983Z^), PM 15682 (M^-^), PM 15866 (M^-^), from Hawk and
Fault localities, Bridger Fm., Green River Basin, Wyoming (see West, 1973); UW 984,

(P3-M4), from Tabernacle Butte, Bridger Fm., Green River Basin, Wyoming.

Discussion. — lf]\Q systematics of this species were treated elsewhere
(Krishtalka and Stucky, 1983a). P. chesteri is known from late Wa-
satchian to late Bridgerian faunas. Material originally assigned to P.
chesteri from the Clarkforkian and early Wasatchian of the Bighorn
Basin (Bown, 1979; Rose, 1981) was referred above to P. protinnom-
inatus.

Compared to P. elegans, P. pauli, and P. protinnominatus, diagnostic
and derived features on the type of P. chesteri, an M3, include an
occlusally narrow talonid (the hypoconid is more internal), and an
absence of the labial emargination between the trigonid and talonid.
These characters are also preserved on UW 984, a partial right dentary
with P3~M4 from Tabernacle Butte (see McGrew, 1959; Bown, 1982),
and a lower molar from Powder Wash. Upper molars from Powder
Wash, the Wind River Formation (Krishtalka and Stucky, 1983a,
1983/?), and the Bridger Formation (West, 1973, PI. 1, Figs. C, D) also
bear distinctive features and are referred to P. chesteri. The morphology
of these specimens indicates that this species is intermediate in molar
structure between P. protinnominatus and P. minutus. P. chesteri has
the basic features of the former, but is more derived in that: there is a
greater disparity in L/W ratio from M* to M^, so that M* is longer than
wide, M^ is more nearly equilateral, and M^ is very transverse; the
paracone and protocone are relatively lower and higher, respectively,
and the paracone and metacone are closer to one another; the conules
and stylar cusps C and D are vestigial or absent, and the protocone is

250

Annals of Carnegie Museum

VOL. 52

<

Fig. 5.—Peradectes chested. (A) UW 984, RP3-M4, approx. X 14; (B) USNM 19199, RM3, type, approx. X 17.

1983

Krishtalka and Stucky— Paleogene and Eocene Marsupials

251

more compressed anteroposteriorly on M^; the lower molars are nar-
rower in proportion to length, the talonids are narrower occlusally, and
the buccal part of the base between the trigonids and talonids is not
emarginate. These features are further modified in P. minutus: the
protocone on is more compressed anteroposteriorly and the para-
cone is reduced in height to that of the protocone and almost to that
of stylar cusp B. On the type of P. chesteri a large gap occurs between
the alveolus for the posterior root of M4 and the ascending ramus of
the dentary. This gap is absent from lower jaws of P. minutus, indicating
a shortening of the jaw in the latter. P. chesteri is divergent from P.
calif ornicus in having a more compressed protocone on M^, greater
disparity in L/W ratio from M* to M^, and in lacking conules and stylar
cusps C and D. Their inferred common ancestry was discussed above.

Peradectes sp. cf. P, minutus

Nanodelphys s,p. nov. Setoguchi, 1973.

Nanodelphys cf N. minutus Setoguchi, 1975, in part.

Referred specimens. -CM 15091, 15628, 15687, 51369 (all M‘), CM 16007, 15094
(both M2), CM 15047, 15618 (both MT CM 18216 (DP3), CM 52415 (M,), CM 15016,
15006, 15087, 15082, 29484, 52413, 52416 (all M2), CM 15666, 15685, 23843, 23847,
52418 (all M3), from localities 5, 5 A, 5 Front, 5 Back, 6, Wood, 20, Badwater Creek
area, Wagon Bed Fm., Wind River Basin, Wyoming.

Discussion. — SeXoguchi (1973) was correct in recognizing the occur-
rence of two peradectines at the late Eocene Badwater localities. For
reasons not given, he later (1975) emended this conclusion and com-
bined the two species as Nanodelphys cf. N. minutus. The latter is a
composite of two taxa, one of which is P. californicus (see above). The
remaining part of the sample represents a primitive variant of P. mi-
nutus, which is here given tentative taxonomic recognition. The few
M“S and M^s are somewhat less transverse than those of P. minutus
and have a somewhat higher paracone in relation to the protocone and
stylar cusp B. Also, parts of the dentition that preserve other diagnostic
features of P. minutus (the short talonid on M4; loss of the gap between
M4 and the ascending ramus of the dentary) are not represented in the
Badwater sample. As such, it seems prudent to distinguish this material
from P. minutus until more conclusive evidence for referral to the
latter is available.

Peradectes sp. cf. P. minutus occurs with P. californicus at five of
the Badwater localities. Teeth of the former are smaller, are more
transverse and have more compressed protocones, and, as a result, the
disparity in L/W ratio from is greater. Also, Mi_3 are narrower
in proportion to length. Although these two species differ in size at
Badwater, their combined range in size is encompassed by the samples
of P. californicus from the Uintan of California. As discussed above.

252

Annals of Carnegie Museum

VOL. 52

this may be an example of character displacement in lithosympatric
species, but larger samples are needed to test this inference.

Peradectes minutus (McGrew, 1937)

When McGrew (1937) named the Orellan Nanodelphys minutus the
Tiffanian P. elegans was the only known species of Peradectes. Large
differences in size and morphology implied generic distinction. How-
ever, the morphology of other species of Peradectes recovered since
then from Paleocene and Eocene horizons indicates that TV. minutus is
the most derived species of a branching lineage that includes P. pauli,
P. protinnominatus, and P. chesteri (and Peradectes sp. cf. P. minutus).
Its advanced dentition appears to be the result of incremental and
cumulative changes in that lineage, and generic distinction is no longer
warranted. Crochet (1978) and Bown (1979) reached a similar conclu-
sion.

Earlier suggestions by Galbreath (1953) and Setoguchi (1973) that
“TV.” minutus may be conspecific with Peratherium huntii are in error.
Cope ( 1 884) and Galbreath (1953) noted the large size of the entoconid
on the type of P. huntii— 2i diagnostic feature of the Didelphini — and,
partly on this basis. Hough (1961) allocated P. huntii to Herpetotherium
fugax.

P. minutus is most closely related to P. chesteri. However, and
especially M- are more transverse, with a more highly compressed
protocone. On M'“^ the reduced paracone and the protocone are sub-
equal in height and slightly higher than an enlarged stylar cusp B (see
McGrew, 1937, Fig. 3; Setoguchi, 1978, Fig. 7A). P. minutus 2l\so lacks
the gap on the mandible between M4 and the ascending ramus that is
evident on the type of P. chesteri (Fig. 5).

P. minutus is known from Oligocene horizons of the western interior
of North America. Late Eocene material that Setoguchi (1975) referred
to Nanodelphys cf. TV. minutus was assigned above to Peradectes sp.
cf. P. minutus, P. californicus, and Peratherium innominatum. As dis-
cussed above, recognition of Peradectes sp. cf. P. minutus is tentative.
It has most of the derived features of P. minutus and, with recovery
of additional material, may prove to be conspecihc with the latter.

Setoguchi’s (1978) sample of "^Nanodelphys new species” from the
Orellan of Badwater is also a composite of numerous taxa. Much of
the material represents a small species of Peratherium or Herpeto-
therium, which explains his attribution of dilambdodonty to this species.
Of the specimens he referred, only CM 19802, 19804, 21697, 33557,
33570, 33574, 33577, 33582, 33589, 33590, 33595, 33598, 33600,
33613, 33619, 33621, 33625, 33628, 33631, 33632, 33642, 33647
belong to the Peradectini, and specifically P. minutus. To these should
also be added CM 33599 (M’-^) and KU 16611 (M^-^).

1983

Krishtalka and Stucky— Paleogene and Eocene Marsupials

253

Armintodelphys Krishtalka and Stucky, 1983

The two species in this genus— /t. blacki and A. dawsoni—arQ known
from the late Wasatchian to early Bridgerian and from the early Brid-
gerian, respectively, of the Wind River Formation. A. dawsoni is also
known from the early Bridgerian Powder Wash locality, Utah (Krish-
talka and Stucky, 1983/?). The systematics of the two species were
treated in detail elsewhere (Krishtalka and Stucky, 1983(2). Briefly,
Armintodelphys is derived over other North American Peradectini in
having M2_4 with a reduced, nubbin-like entoconid that is smaller and
lower than the hypoconulid, and M2 with a narrower talonid than
trigonid. Otherwise, the two species most closely resemble P. pauli in
retaining high molar trigonids, long talonids, a buccal emargination
between the trigonids and talonids, a longer talonid than trigonid on
M4, and a narrower talonid than trigonid on M3. More derived species
of Peradectes have an M4 with a shorter talonid than trigonid, as well
as other advanced features. Accordingly, Armintodelphys may have
evolved from or shared a common ancestry with P. pauli.

An M^ from Powder Wash, tentatively referred to A. dawsoni (Krish-
talka and Stucky, 1983/?), differs from that of known species of Pera-
dectes in having a weak, compressed protocone, an enlarged stylar cusp
C, more closely approximated paracone and metacone, and expanded
stylar salients. As in other Peradectini, the M- is not dilambdodont.

Mimoperadecles Bown and Rose, 1979

The only described species in this genus, M. labrus, occurs in early
Wasatchian faunas from the Bighorn and Powder River basins. Apart
from being the largest known species in the Peradectini, M. labrus is
specialized in having a paraconid on M2_4 that is larger and higher than
the metaconid. It shares with species of Peradectes the morphology of
a low, subequal, and closely appressed entoconid and hypoconulid on
the lower molars. It differs from all species of Peradectes except P.
elegans in retaining an Mj that is as large as M2, an M^ that is not
transverse, and M““^ with comparatively strong conules and stylar cusps
C and D, and an uncompressed protocone. As such, Mimoperadectes
most closely resembles P. elegans, but is more derived in having an
M3 with a narrower talonid than trigonid. These features imply that
Mimoperadectes originated in common with P. pauli from a Paleocene
species of Peradectes that had developed the narrow M3 talonid. As
Bown and Rose (1979) noted, the enlarged paraconid on the lower
molars is convergent on the condition in stagodontid marsupials. Un-
like the latter, M. labrus lacks the enlarged, specialized posterior pre-
molars, and compressed trigonids on the lower molars.

254

Annals of Carnegie Museum

VOL. 52

Evolutionary Relationships

The depiction of relationships (Fig. 6) among Peratherium, Herpe-
totherium, Peradectes, Mimoperadectes, and Annintodelphys is based
on current thought (Clemens, 1966, 1968, 1979; Fox, 1979) that an
Alphadon-XikQ dentition represents the primitive condition among di-
delphines. Accordingly, the following molar characters are primitive
for Tertiary North American didelphines: on Mi_3, the talonid is equal
in width to, but much longer than, the trigonid; the cristid obliqua
meets the posterior wall of the trigonid medially, below the ventral
apex of the protocristid; the entoconid is high and conical; the hypo-
conulid is subequal or lower than, and posterior or posterolabial to,
the entoconid; the entoconid notch is moderate or large; the metaconid
is larger than the paraconid; on M4, the talonid is much longer than
the trigonid; on the paracone and metacone are subequal; the

centrocrista is straight (not dilambdodont) or weakly directed toward
the stylar shelf (incipiently dilambdodont); the stylar cusps and conules
are well developed.

In comparison. Tertiary North American didelphines are advanced
dentally in having a shortened talonid on M,_3 and a more labial cristid
obliqua on the lower molars (Fig. 6, node 1)~ features that imply their
common ancestry. Dental characters among these genera suggest a
subsequent divergent radiation: on the one hand, members of the Di-
dAphini— Peratherium and Herpetotherium — retain one variant of the
primitive Alphadon-like hypoconulid-entoconid complex, but are spe-
cialized in having dilambdodont with a reduced paracone in

comparison to the metacone and a reduced ectoflexus (Fig. 6, node 2).
On the other hand, members of the Feradectini— Peradectes, Mimo-
peradectes, Annintodelphys— appear united by a different suite of de-
rived features: on Mi_3, a reduced entoconid and hypoconulid that are
closely appressed, subequal, and separated by a weak entoconid notch,
and arise from a common internal talonid wall; on reduced stylar
cusps and conules (Fig. 6, node 3). These genera retain the primitive,
non-dilambdodont, structure of the upper molars.

Although the common ancestry of the Didelphini and Peradectini is
implied by their possession of a shorter talonid and a labial cristid
obliqua on Mi_3, their origin is difficult to decipher. The derived char-
acters of each group appear to be variable among the species of Al-
phadon (see Clemens, 1966; Lillegraven, 1969; Fox, 1971, 1979). For
example, published figures of the dentition indicate that dilambdodonty
does not occur among species of Alphadon. However, incipient di-
lambdodonty, where the centrocrista or just the premetacrista is di-
rected toward the stylar shelf, is evident on some upper molars of A.
marshi, A. lulli, A. rhaister, and A. russelli. Some species of Alphadon

1983

Krishtalka and Stucky— Paleogene and Eocene Marsupials

255

Fig. 6. — Relationships among Paleocene and Eocene North American marsupials. Node
1: shorter talonid on M,_3; cristid obliqua meets trigonid labially, below protoconid on
Mj_3; Node 2: dilambdodont, paracone much lower than metacone and ectoflexus

reduced; on Mi_3, entoconid taller and spire-like, hypoconulid much lower and posterior,
entoconid notch veiy- deep; Node 3: on M,_3, reduced entoconid and hypoconulid sub-
equal, closely appressed and share common internal talonid wall; entoconid notch weak;
on stylar cusps C and D and conules reduced; Node 4: M3 talonid narrower (basally)
than trigonid; Node 4A: large size, M2_4 paraconid larger and higher than metaconid;
Node 5: Mj smaller than M2; more transverse, with moderately compressed pro-
tocone; Node 5A: M2_4 entoconid and entoconid notch vestigial; M2 talonid narrower
(basally) than trigonid; Node 6: M4 talonid shorter than trigonid; P3 with shorter talonid;
P3 lower than M,; protocone more anterior on M*-^; Node 7: Mi_3 talonids with narrower
occlusal width (hypoconid more internal, cristid obliqua runs posteriorly from trigonid);

with reduced conules and stylar cusps C and D; Node 7A: P3 talonid lost; Node 8:
greater disparity in L/W ratio from to M^; conules and stylar cusps C and D vestigial;

more transverse with a more highly compressed protocone; loss of labial emargination
between trigonid and talonid on Mi_3; Mj_3 narrower in proportion to length; Node 8A:
M*-^ paracone reduced to height of protocone and almost to that of enlarged stylar cusp
B; and especially M^ more transverse, with a more highly compressed protocone;
loss of gap between M4 and ascending ramus (shortening of jaw).

appear to have a tall entoconid, a lower and posterior hypoconulid,
and a strong entoconid notch on some of the lower molars. Others
have a subequal entoconid and hypoconulid and a weak entoconid
notch. Compared to other species of Alphadon, A. russelli (Fox, 1979)
seems to most closely resemble the Peradectini in having M2_3 with
short talonids, a labial cristid obliqua, closely appressed, low and sub-

256

Annals of Carnegie Museum

VOL. 52

equal entoconid and hypoconulid, and a weak entoconid notch. Sim-
ilarly, some specimens of A. rhaister (see Clemens, 1 966) and A. rnarshi
(see Lillegraven, 1969) show affinity to the Didelphini in the greater
development of dilambdodonty.

These observations cannot, as yet, suggest specific evolutionary re-
lationships. The distribution of these features within and among species
of Alphadon have not been emphasized, nor, perhaps, found taxonom-
ically useful by students of Cretaceous didelphines. Conversely, diag-
nostic features cited for Cretaceous didelphines have not proven useful
in distinguishing among the Tertiary ones. Given this state of affairs,
the only evolutionary conclusion warranted at this time is that the
species of Alphadon display a range of morphology that encompasses
some of the distinctive features of Tertiary Peradectini and Didelphini.
The divergence of the two tribes will be better understood when the
distribution of these features among the species of Alphadon are doc-
umented.

One corollary of this conclusion is that Cretaceous didelphines, spe-
cifically Alphadon and Albertatheriurn, cannot be included in the Per-
adectini (contra Crochet, 1979) or Didelphini; none of the Cretaceous
species has, as yet, been demonstrated to have the suite of diagnostic
and derived features of either tribe. A second is that Thylacodon pus-
illus may not be a species of Peradectes nor the Peradectini, if figures
of Peradectes cf P. pusillus (Archibald, 1982, Figs. 42, 43), from the
Puercan of Montana, accurately represent the morphology of the species.
They indicate that M‘ and have large stylar cusps C and D, and
Mi_3 have elongated talonids, very high trigonids, large entoconids, i
and deep entoconid notches. None of the figured specimens exhibits
the derived features of Peradectes or the Peradectini. A third corollary
is that the common supposition of an ancestral-descendant relationship
between Peradectes and Peratherium (Clemens, 1968) is unsupport-
able; at its earliest appearance Peradectes (ss) is too derived in features
of the lower dentition.

Fig. 7.— Occurrences of North American Eocene species of Peratherium. Distance be-
tween ages and subages (GB — Graybullian; LY— Lysitean; LC— Lostcabinian; GA—
Gardnerbuttean), and placement of occurrences (black dots) are diagrammatic. Letter
abbreviations denote areas of occurrences; numbers refer to original sources for identi-
fications and occurrences. AF— Agua Fria; BHB— Bighorn Basin; BRB — Bridger Basin;
BSB— Bison Basin; CA— California; FMA— Four Mile area; GRB— Green River Basin;
HB — Huerfano Basin; PRB— Powder River Basin; SJB— San Juan Basin; SWB— Sand
Wash Basin; WB— Washakie Basin; WRB— Wind River Basin; UB — Uinta Basin. 1 —
this paper, Krishtalka and Stucky, 1983«2; 2~Bown, 1979; 3— -Bown, 1982; 4 — Bown
and Rose, 1979; 5-Delson, 1971; 6--Eaton, 1982; 7-Gazin, 1952; 8-Gazin, 1956;

1983

Krishtalka and Stuck y— Paleogene and Eocene Marsupials

257

z

m

LU

X

o

3

Q

1

1

1

|^WRB22

1

1

!

I

I

UINTAN

twRB^* “

!ca^®

^WRB22

1

!

!

1

|| OiAfpSO

1

>F31

Q

1

O

|brb‘*°

1

qrb10,15

i

j

I

I

1

f

Abrb^

i

1

BRIDGERIAN

la

<

1

Ibhb®

JuB«

BRB^O’23

1

i

I

!qrb1.28

I brb"'®

!

!

1

I

iBHB®

IBrb10,18,27?

!GRB2®

jUB^2

!

1

!

1

1

i

1

1

f

O

1

!

I

!

I

!

^wrb’’

I

1

'■'■f '

1

I?HB25

I

S

\

1

^wrb'*

I

I

!

1

iwRB^ •GRB.^

1

1

1

iwRB^

i

!

I

I

1

J P. edwardi

J

1

z

<

■ ! — —

1

I

j

1

1

!

1

I

o

%

4wrb^

1

1

!

!

!

iwRB^

P. marsupium

4sjb^^

!

1

m

I

!

!

AbHB21

o

P. innominatum

P. knighti

P. comstocki

9— Gazin, 1962; 10— Gazin, 1976; 11— Guthrie, 1971; 12— Krishtalka and Stucky, 19836;
13 — Lillegraven, 1976; 14— Lucas et al., 1981; 15 — Marsh, 1872; 16 — Matthew and
Granger, 1921; 17— McGrew, 1959; 18 — McGrewand Sullivan, 1970; 19 — McKenna,
1960; 20— Rose, 1981; 21 — Schankler, 1980; 22 — Setoguchi, 1973, 1975; 23 — Simpson,
1928; 24 — Simpson, 1935; 25 — Simpson, 1968; 26 — Stock, 1936; 27— Troxell, 1923;
28— West, 1973; 29— West and Dawson, 1973; 30— West and Dawson, 1975; 31 —West,
1982.

258

Annals of Carnegie Museum

VOL. 52

The five known Eocene species of Peratherium in North America
differ principally in size (Tables 1 and 3), and bear the diagnostic
trademarks of the genus at their earliest appearance. Thus, their pre-
served morphology does not provide any clues to the relationships
among the five species. Peratherium is unknown from pre-Wasatchian
faunas in North America (Fig. 7). Three species {P. comstocki, P. knigh-
ti, P. innominatum) appear penecontemporaneously in the early Wa-
satchian. This may imply that Peratherium immigrated to the western
interior of North America near the onset of the Wasatchian. If so, this
may have involved one species, with subsequent diversification, or a
multiple immigration.

Three of the five species of Peratherium (P. comstocki, P. marsu-
pium, P. knight i) show no detectable morphologic change throughout
their stratigraphic extent, which according to West et al. (manuscript),
corresponds to approximately 7, 9, and 1 1 million years, respectively.

P. innominatum shows a gradual and slight anagenetic increase in size
from the early Wasatchian to the Duchesnean (Krishtalka and Stucky,
1983(2), a period of approximately 11 million years. This long-term
stasis in the preserved dental morphology of these species is in sharp
contrast to the evolutionary tempo and mode of North American species
of Peradectes, as well as many placentals, during this period of time.

The six known North American Tertiary species of Peradectes differ
in aspects of dental morphology other than size. P. elegans (Fig. 6,
node 3) exhibits the primitive features of the Peradectini. P. pauli and
remaining species of Peradectes (Fig. 6, node 5) are more derived in
having a smaller Mj than M2, an M3 with a narrower talonid than
trigonid, and a more transverse M^ with a compressed protocone. P.
protinnominatus and remaining species (Fig. 6, node 6) are advanced
over P. pauli in having a shorter talonid than trigonid on M4, a reduced
P3, and a more anterior protocone on M‘“^. P. californicus, P. chesteri,
and P. minutus (Fig. 6, node 7) are further derived in having reduced
conules and stylar cusps C and D on M*"^, and narrower talonids
occlusally on Mi_3. P. chesteri and P. minutus (Fig. 6, node 8) have a
greater disparity in the L/W ratio from M^ to M^, more compressed
protocones, vestigial or absent stylar cusps C and D and conules, nar- [
rower lower molars in proportion to length, and no emargination be-
tween the molar trigonids and talonids. P. minutus (Fig. 6, node 8A)
seems most advanced in the hypertrophy of stylar cusp B, the reduction
in height of the paracone, and the severe compression of the protocone
on the upper molars.

In summary, the primitive features of the Peradectini, expressed in
P. elegans, are modified in P. pauli and, in turn, in P. protinnominatus,

P. californicus, P. chesteri, and P. minutus. The major and discrete
modifications are: reduction of P3, Mj, the molar talonids, and the

1983

Krishtalka and Stucky — Paleogene and Eocene Marsupials

259

conules and stylar cusps C and D; compression of the protocone on
the upper molars; and an increase in the disparity of L/W ratio from
M* to M^. Other modifications are less discrete. They are variable
within a species to the point of near overlap with another species. For
example, the progressive reduction in the buccal emargination on the
lower molars, in the relative height of the trigonids, and in the height
of the paracones on the upper molars appear to be part of a morpho-
cline. They are valuable as diagnostic criteria (and are only used as
such) in species that express disparate parts of that morphocline.

When the reconstruction of relationships among the species of Per-
adectes is superimposed on their stratigraphic record (Fig. 8), the po-
larity of morphologic change is consistent with the increasingly younger
temporal occurrence of these species. Accordingly, these species appear
to form a branching lineage that is characterized by speciation and
gradual morphologic change. An ancestral-descendent relationship is
implied in the P. pauli-P. protinnominatus lineage, with subsequent
cladogenetic divergence of a P. chesteri-P. minutus lineage and P. cal-
ifornicus. Juxtaposition of the stratigraphic record and dental mor-
phology of these species implies that each of the lineages is characterized
by incremental and gradual morphologic change, and the appearance,
anagenetically, of new, discrete phena, here termed species. None of
these species has uniquely derived features that would preclude such
a phylogenetic reconstruction. None of these lineages form an evolu-
tionary species (sensu Simpson, 1961). To combine the species within
them as such would deny the current paleontological evidence of the
appearance of discrete and incremental evolutionary novelties in a
temporal succession of progressively younger species. Gaps in the geo-
logic record (and in our knowledge) of these species between their
successive first and last occurrences are not evidence against the in-
ferred lineages, nor evidence for an alternative interpretation. When
those gaps are filled, that new evidence will lend support, demand
modification, or refute the lineages inferred from current evidence.

Armintodelphys (Fig. 6, node 5A) may have evolved in common with
P. pauli\ it has the derived features of the latter, but is specialized in
having a vestigial entoconid on M2_4 and a narrower talonid than tri-
gonid on M2. The affinities of Mimoperadectes are less clear. Except
for an M3 with a narrower talonid than trigonid, the dentition of M.
labrus (Fig. 6, node 4A) most closely resembles that of P. elegans, and
lacks the derived features of other species of Pemdectes. It may have
evolved in common with P. pauli from a Paleocene species of Pera-
dectes with a reduced talonid on M3. M. labrus is specialized in its large
size, and in having a larger paraconid than metaconid on M2_4.

The fossil record of North American early Tertiary didelphines re-
veals contrasting evolutionary histories. Species of the Didelphini ap-

260

Annals of Carnegie Museum

VOL. 52

WRB^

CA13, 26

WRB^-22
P. californicus

f WRB^’ 22
I

1 -

I

4 WRB"'' 22
Peradectes sp.
cf. P. minutus

f GRB^-’’^

I

j

’ UB ^’■'2

*GRB28

I

I

4wrb^

I

1 — — __

I

*GRB^

P. chesteri

•UB^2

I

I

I

^WRB^ iwRB
I A. dawsoni

1

I

• WRB^

A. blacki

QQ

O

• FMA1-19

FMA19

tBHB^’2

I

I

• BHBI-20
P. protinnominatus

f SJB16'24 ~

4bsb8 •BSB8

P. elegans P. pauli

BHB^

M. labrus PRB'^’S

Fig. 8. — Occurrences of North American Paleocene and Eocene species of Peradectini.
For explanation of symbols see legend for Fig. 7.

1983

Krishtalka and Stucky— Paleogene and Eocene Marsupials

261

pear abruptly and penecontemporaneously, and remain virtually un-
changed throughout their temporal extent. Among the Peradectini,
species of Peradectes appear sequentially, exhibit cumulative modifi-
cations, and apparently form a branching lineage that is characterized
by anagenesis and cladogenesis. Armintodelphys and Mimoperadectes,
are offshoots of that lineage.

Acknowledgments

We thank Mary R. Dawson for discussions concerning the early Tertiary didelphine
marsupials and for reviewing the manuscript, and Richard C. Fox for sharing an un-
published manuscript on Herpetotherium. The following kindly provided access to mar-
supial material in their care: Donald Baird (PU), Robert Emry (USNM), Philip Gingerich
(UM), Howard Hutchison (UCMP), Malcolm McKenna (AMNH), John Ostrom (YPM)
and Peter Robinson (UCM). We are grateful to Nick Piesco (University of Pittsburgh,
School of Dentistry) for use of the SEM facilities and for his technical assistance. This
work was supported in part by the Rea Postdoctoral Fellowship, Carnegie Museum of
Natural History.

Literature Cited

Archibald, J. D. 1982. A study of Mammalia and geology across the Cretaceous-
Tertiary boundary in Garheld County, Montana. Univ. California Publ. Geol. Sci.,
122:1-286.

Aymard, a. 1 846. Essai Monographique sur un genre nouveau de mammiferes fossile
trouve dans le Haute-Loire et nomme Entelodon. Ann. Soc. Agr. du Puy, 12:227-
267.

. 1850. Compte rendu de la seance du 13 avril 1849, reponse a M. Robert sur

les mammiferes fossiles des calcaires du Puy. Ann. Soc. Agr. du Puy, 14:80-86.
Bown, T. M. 1979. Geology and mammalian paleontology of the Sand Creek facies.

Lower Willwood Formation (Lower Eocene), Washakie County, Wyoming. Mem.
Geol. Survey Wyoming, 2:1-151.

. 1982. Geology, paleontology, and correlation of Eocene volcaniclastic rocks,

southeast Absaroka Range, Hot Springs County, Wyoming. Geol. Survey Prof Pap.
1201-A:l-75.

Bown, T. M., and K. D. Rose. 1979. Mimoperadectes, a new marsupial, and Worlandia,
a new dermopteran, from the lower part of the Willwood Formation (Early Eocene),
Bighorn Basin, Wyoming. Univ. Michigan, Contribs. Mus. Paleont., 25:89-104.
Clemens, W. A. 1966. Fossil mammals of the type Lance Formation, Wyoming. Part
IF Marsupialia. Univ. California Publ. Geol. Sci., 62:1-122.

. 1968. Origin and early evolution of marsupials. Evolution, 22:1-18.

. 1979. Marsupialia. Pp. 192-220, in Mesozoic mammals, the hrst two-thirds

of mammalian history (J. A. Lillegraven, Z. Kielan-Jaworowska, and W. A. Clemens,
eds.), Univ. California Press, Berkeley, x + 31 1 pp.

Cope, E. D. 1884. The Vertebrata of the Tertiary formations of the West. Report U.S.

Geol. Surv. Terr., F. V. Hayden, Washington, 3:1-1009.

Crochet, J-Y. 1977. Les Didelphidae (Marsupicarnivora, Marsupialia) holarctiques
Tertiares. C. R. Acad. Sci. Paris, 284, serie D:357-360.

. 1978. Les marsupiaux du Tertiaire d’Europe. Acad. Montpellier, Univ. Sci.

Tech. Languedoc, These, 360 pp.

. 1979. Diversite systematique des Didelphidae (Marsupialia) Europeens Ter-
tiares. Geobios, 12:365-378.

262

Annals of Carnegie Museum

VOL. 52

Delson, E. 1971. Fossil mammals of the Wasatchian Powder River local fauna, Eocene
of northeast Wyoming. Bull. Amer. Mus. Nat. Hist., 146:309-364.

Eaton, J. G. 1982. Paleontology and correlation of Eocene volcanic rocks in the Carter
Mountain area. Park County, southeastern Absaroka Range, Wyoming. Univ. Wyo-
ming Contrib. Geol., 21:153-194.

Fox, R. C. 1971. Marsupial mammals from the early Campanian Milk River For-
mation, Alberta, Canada. Zool. J. Linnaean Soc., 50 (Suppl. 1): 145-1 64.

. 1979. Mammals from the Upper Cretaceous Oldman Formation, Alberta. I.

Alphadon Simpson (Marsupialia). Canadian J. Earth Sci., 16:91-102.

Galbreath, E. C. 1953. A contribution to the Tertiary geology and paleontology of
northeastern Colorado. Vertebrata, 4:1-120.

Gazin, C. L. 1952. The Lower Eocene Knight Formation of western Wyoming and its
mammalian faunas. Smithsonian Misc. Coll., 1 17:1-82.

. 1956. Paleocene mammalian faunas of the Bison Basin in south-central Wy-
oming. Smithsonian Misc. Coll., 131:1-57.

— -. 1962. A further study of the Lower Eocene mammal faunas of southwestern

Wyoming. Smithsonian Misc. Coll., 144:1-98.

— . 1976. Mammalian faunal zones of the Bridger middle Eocene. Smithsonian

Contrib. Paleobiol., 26:1-25.

Gray, J. E. 1821. On the natural arrangement of vertebrose animals. London Med.
Reposit., 15:296-310.

Guthrie, D. A. 1971. The mammalian fauna of the Lost Cabin Member, Wind River
Formation (Lower Eocene) of Wyoming. Ann. Carnegie Mus., 43:47-1 13.

Hough, J. R. 1961. Review of Oligocene didelphid marsupials. J. Paleont., 35:218-
228.

Krishtalka, L., and R. K. Stucky. 1983i3. Revision of the Wind River faunas, early
Eocene of central Wyoming. Part 3. Marsupialia. Ann. Carnegie Mus., 52:205-227.

. 1983Z?. Middle Eocene marsupials from northeastern Utah. Ann. Carnegie

Mus., in press.

Lillegraven, j. a. 1969. Latest Cretaceous mammals of upper part of Edmonton
Formation of Alberta, Canada, and review of the marsupial-placental dichotomy in
mammalian evolution. Vertebrata, 50:1-122.

. 1976. Didelphids (Marsupialia) and Uintasorex (?Primates) from later Eocene

sediments of San Diego County, California. Trans. San Diego Soc. Nat. Hist., 18:
85-112.

Lucas, S. G., and R. V. Ingersoll. 1981. Lexicon and bibliography of Cenozoic
deposits of New Mexico. Bull. Geol. Soc. Amer., 92 (Part 2): 1807-1 981.

Lucas, S. G., R. M. Schoch, E. Manning, and C. Tsentas. 1981. The Eocene bio-
stratigraphy of New Mexico. Bull. Geol. Soc. Amer., 92 (Part 1):95 1-967.

Marsh, O. C. 1872. Preliminary description of new Tertiary mammals. Amer. J. Sci.,
ser. 3, 4:202-224.

Matthew, W. D. 1899. A provisional classification of the freshwater Tertiary of the
West. Bull. Amer. Mus. Nat. Hist., 12:19-75.

. \9Q9a. Faunal lists of the Tertiary Mammalia of the West. Bull. U.S. Geol.

Surv., 361:91-138.

. 19097>. The Carnivora and Insectivora of the Bridger Basin, Middle Eocene.

Mem. Amer. Mus. Nat. Hist., 9:291-567.

Matthew, W. D., and W. Granger. 1921. New genera of Paleocene mammals. Amer.
Mus. Novitates, 13:1-7.

McGrew, P. O. 1937. New marsupials from the Tertiary of Nebraska. J. Geol., 45:
448-455.

. 1959. Marsupialia. Pp. 147-148, in The geology and paleontology of the Elk

1983

Krishtalka and Stucky— Paleogene and Eocene Marsupials

263

Mountain and Tabernacle Butte area, Wyoming. Bull. Amer. Mus. Nat. Hist., 1 17:

117-176.

McGrew, P. O., and R. Sullivan. 1970. The stratigraphy and paleontology of Bridger
A. Univ. Wyoming Contrib. Geol., 9:66-85.

McKenna, M. C. 1960. Fossil Mammalia from the early Wasatchian Four Mile local
fauna. Eocene of northwest Colorado. Univ. California Publ. Geol. Sci., 37:1-130.
Robinson, P. 1968. Nyctitheriidae (Mammalia, Insectivora) from the Bridger For-
mation of Wyoming. Univ. Wyoming Contrib. Geol., 7:129-138.

Rose, K. D. 1981. The Clarkforkian Land-Mammal Age and mammalian faunal com-
position across the Paleocene-Eocene boundary. Univ. Michigan Pap. Paleont., 26:
1-197.

ScHANKLER, D. M. 1980. Faunal zonation of the Willwood Formation in the central
Bighorn Basin, Wyoming. Pp. 99-1 14, in Early Cenozoic paleontology and stratig-
raphy of the Bighorn Basin, Wyoming (P. D. Gingerich, ed.), Univ. Michigan Pap.
Paleont., 24:vi + 1-146.

Setoguchi, T. 1973. Late Eocene marsupials and insectivores from the Tepee Trail
Formation, Badwater, Wyoming. Unpublished Ph.D. dissert., Texas Tech Univ.,
Lubbock, 101 pp.

. 1975. Paleontology and geology of the Badwater Creek area, central Wyoming.

Part 1 1. Late Eocene marsupials. Ann. Carnegie Mus., 45:263-275.

—. 1978. Paleontology and geology of the Badwater Creek area, central Wyoming.

Part 16. The Cedar Ridge local fauna (late Oligocene). Bull. Carnegie Mus. Nat.
Hist., 9:1-61.

Simpson, G. G. 1928. American Eocene didelphids. Amer. Mus. Novitates, 303:1-7.

. 1935. The Tiffany fauna. Upper Paleocene. I. — Multituberculata, Marsupialia,

Insectivora, and ?Chiroptera. Amer. Mus. Novitates, 795:1-19.

. 1961. Principles of animal taxonomy. Columbia Univ. Press, New York,

247 pp.

. 1968. A didelphid (Marsupialia) from the early Eocene of Colorado. Postilla,

115:1-3.

Stock, C. 1936. Sespe Eocene didelphids. Proc. Nat. Acad. Sci., 22:122-124.
Troxell, E. L. 1923. A new marsupial. Amer. J. Sci., 5:507-510.

West, R. M. 1973. Geology and mammalian paleontology of the New Fork-Big Sandy
area, Sublette County, Wyoming. Fieldiana, 29:1-193.

. 1982. Fossil mammals from the Lower Buck Hill Group, Eocene of Trans-

Pecos Texas: Marsupicarnivora, Primates, Taeniodonta, Condylarthra, bunodont
Artiodactyla, and Dinocerata. Texas Mem. Mus. Pearce-Sellards Ser., 35:1-20.
West, R. M., and M. R. Dawson. 1973. Fossil mammals from the upper part of the
Cathedral Bluffs tongue of the Wasatch Formation (early Bridgerian), northern Green
River Basin, Wyoming. Univ. Wyoming Contrib. Geol., 12:33-41.

. 1975. Eocene fossil Mammalia from the Sand Wash Basin, northwestern Moffat

County, Colorado. Ann. Carnegie Mus., 45:231-253.

West, R. M. et al. manuscript. Eocene (Clarkforkian through Duchesnean) chronology
of North America.

WoRTMAN, J. L. 1901. Studies of Eocene Mammalia in the Marsh Collection, Peabody
Museum. Part 1. Carnivora. Amer. J. Sci., ser. 4, 1 1:333-348.

Back issues of many Annals of Carnegie Museum articles are
available, and a few early complete volumes and parts are listed
at half price. Orders and inquiries should be addressed to: Pub-
lications Secretary, Carnegie Museum, 4400 Forbes Avenue,
Pittsburgh, Pa. 15213.

7 73

ISSN 0097-4463

ANNALS

0/ CARNEGIE MUSEUM

CARNEGIE MUSEUM OF NATURAL HISTORY

4400 FORBES AVENUE * PITTSBURGH, PENNSYLVANIA 15213

VOLUME 52 16 SEPTEMBER 1983 ARTICLE 11

THE CHIRA BEACH RIDGES, SEA LEVEL CHANGE, AND
THE ORIGINS OF MARITIME ECONOMIES ON
THE PERUVIAN COAST

James B. Richardson III

Chief Curator, Section of Man

Abstract

The Chira beach ridges have been radiocarbon dated and provide the best available
evidence for the establishment of the north Peruvian coast at 5000 B.P. The stabilization
of sea level coincides with a change in the east Pacific Ocean current patterns. These two
natural events are seen as major causal factors in the subsequent rise of complex pre-
ceramic maritime societies on the coast of Peru.

Introduction

The question of the origins and development of maritime societies
on the Peruvian coast has engendered heated discussions by the ar-
chaeological community. However, little attention has been paid by
archaeologists to the various disciplines, especially oceanography, that
are able to provide many of the answers that have been posed con-
cerning this crucial juncture in Peruvian cultural development. Chang-
ing environments and biological systems are key to our understanding
of man’s adaptation to this coast through time. Geological, tectonic,
environmental, and oceanographic changes and the El Nino perturba-
tion, all must be an integral part of any discussion centering upon the
emergence of Peruvian maritime societies.

There are two environmental factors that are crucial to any discussion
concerning the emergence of maritime based societies on the Peruvian
coast: the date of the establishment of the modern climate and the

Submitted 16 October 1981.

265

266

Annals of Carnegie Museum

VOL. 52

formation of the present coastline. A major change in the east Pacific
current patterns takes place prior to 5000 B.P., which, in turn, is re-
sponsible for the northward shift the cold Peru current that brings with
it the major fishery that was intensively exploited by late preceramic
populations (Rollins et ah, 1981; Sandweiss et ah, 1983). This event
coincides with the formation of the present coastline of Peru at the
time when modern sea level was reached. The beach ridges of the Chira
River of northwest Peru, provide the best information of the date when
modern sea level was attained. Both these two events occurred at circa
5000 B.P. and it is after this date that maritime oriented societies
develop on the coast.

Holocene Environmental Change

In 1981, I presented a model to explain the emergence of maritime
societies on the Peruvian coast (Richardson, 1981). A main causal
factor was seen to be the rise of sea level to present levels by 5000 B.P.
It was demonstrated that the use of maritime resources can be traced
as far back as 10,000 B.P. and that the evidence for early maritime
exploitation lies submerged on the continental shelf, because the Pe-
ruvian coast was many times wider during the Holocene. Complex
maritime societies were established when modern sea level and essen-
tially modern distributions of molluscan and fish resources was at-
tained. The developments leading up to the establishment of sedentary
maritime societies on the Peruvian coast is to be found on the conti-
nental shelf, for the origin of maritime subsistence patterns was not an
abrupt transition from hunting and gathering to littoral resources as so
often stated, but rather a gradual process of adaptation that preceded
the pre-cotton preceramic by several thousand years.

Prior to 5000 B.P., the coastal environment was also markedly dif-
ferent from that of the present. Campbell (1982) reconstructs the en-
vironment of northwestern Peru as one with annual monsoon rains,
savannas, forests, lakes, and marshlands at 14,000 B.P. The mangrove
molluscan fauna from archaeological sites, north of Lambayeque point
to continued wetter conditions as late as 5500 B.P. (Richardson, 1978;
Cardenas, 1979).

Recent studies in the Santa Valley region has resulted in the iden-
tification of an extensive warm water molluscan fauna in a series of
preceramic sites (Rollins et al., 1981; Sandweiss et al., 1983). This
fauna suggests that the ocean current system of the eastern Pacific was
different from that of today. Dating to 5000 B.P., these shellfish species
now inhabit the warm water Panamic-Province north of 7.5° south
latitude. This study and corroborating evidence from marine sediments
off the Central Peruvian Coast also points to a warm water regime
prior to 5000 B.P. (DeVries, 1979).

1983

Richardson —Peruvian Maritime Economies

267

This ever-growing body of evidence now suggests that we can only
extend our present day model of the current pattern of this portion of
the east Pacific Ocean back about 5000 B.P. The evidence strongly
suggests that from at least 1 1,000 B.P. to about 5000 B.P., the central
and northern coasts of Peru were bathed by warm Panamic Province-
type waters, which today are found only north of Paita, about 5° south
latitude. If the Equatorial warm water current was a yearly visitor to
the coast or a year around phenomena, it would have major im.plica-
tions for a different climate and resource base available to pre-5000
B.P. coastal populations. Rainfall would have fallen on coast seasonally
and land based resources may have been more abundant. If this senario
is valid, the coast of Peru, from Lima northward, would have had a
warm water current bathing its shores. The northern extension of the
Peru Current with its upwelling and its abundance of resources would
have been south of Lima, moving northward and becoming established
in its present position after 5000 B.P. The hunters, gatherers, and
maritime resource users would have enjoyed a higher abundance of
land based resources in grasslands/forests across the wide expanse of
the continental shelf prior to 5000 B.P. The warm water ocean re-
sources, may have been markedly less than those enjoyed by post- 5 000
B.P. populations subsisting on cold water faunas. The dramatic change
from, the warm water to a cold water ocean regime prior to 5000 B.P.
decreased rainfall on the coast to the point of turning the landscape
into its current desert environment. The implacement of the cold water
current by 5000 B.P. across the widest expanse of the continental shelf
betw'een 6° and 1 5° south latitude, provided the resource base for the
succeeding rise of complex maritime societies on the Peruvian coast.
The modern weather patterns of the eastern Pacific was established by
5000 B.P. and at the same time, sea level attained its present position.
It is with the establishment of the modern coastline and climate that
large archaeological sites appear in profusion in this area of the Pe-
ruvian coast.

The Chir.a Beach Ridges

The best evidence to date for the establishment of the modern coast-
line on the western South American coast are the beach ridges of the
Peruvian north coast. There are three sets of Holocene ridges, one
situated north of the mouth of the Piura River and sets south and north
of the Chira River (Fig. 1). Only the set north of the Chira River will
be presented in detail.

The research into the Pleistocene marine terraces and Holocene Beach
ridges has a long history in northwest Peru, stemming back to Thomas
Bosworth’s Geology of the Tertiary and Quaternary Periods in the North-
West Part of Peru (1922) in which he discussed the Pleistocene tablazo

268

Annals of Carnegie Museum

VOL. 52

83° 82° 81° 80°

Fig. 1. — Location map of the Chira and Piura River beach ridges.

- 5°

79°

systems and the Holocene deposits of the Talara area. Other investi-
gators, notably Horace Richards (1962), Lemon and Churcher (1961), j
Chotowski (manuscript), Woodman and Folia (1974), Nestor Chigne j
(1975), and Campbell (1982) have made studies of the Pleistocene and !
Holocene deposits. The archaeological research on the occupation of !
the Chira ridges began in 1965 in conjunction with a major research
program on climate change and cultural development in the Chira and ;
Piura River valleys. The twelve unpublished radiocarbon dates from
these beach ridges provide the best sequence of Holocene beach ridge
succession in Peru.

There are three raised Pleistocene marine floors fronting the Amo-
tape Mountains— the Mancora (61-330 m high), the Talara (49-90 m

1983

Richardson— Peruvian Maritime Economies

269

high) and the latest, the Lobitos Tablazo (14-41 m high). These former
marine floors overlie a truncated surface of the Cenozoic and the rocks
of the Pennsylvania Amotape formation and are composed of a thick
sequence of marine quartz sands, shelly and calcareous sands, marls,
coquinas, and pebble beds. Each of these tablazos represent a period
of marine encroachment, followed by vertical uplifts along the Pacific
fault line, bordering the continental shelf.

Of interest to this discussion is the Lobitos Tablazo, little of which
remains at present. The main section, still extant, is a thin ridge fronting
the Holocene beaches acting as a barrier behind which mangrove vege-
tation, during the late Pleistocene and early Holocene, thrived (Rich-
ardson, 1978). The Lobitos ridge consists of a wave cut beach on its
western margin. The mollusks of the western edge, facing the ocean
are salt water species, whereas the mollusks on the eastern face consist
of mangrove and other warm species reflecting a shallow lagoonal
environment.

The nine beach ridges emanate out of the Chira River and stretch
30 km to Punta Balconas and Punta Parinas at the modern town of
Negritos. Parallel to the Lobitoas Tablazo, there are nine Holocene
beach ridges which can be divided into two groups. The first four, with
sharply defined profiles, are separated by broad swales and which are
easily discernable, whereas the second group of five, presents a complex
series of hummocky ridges with intermittent swales upon which there
are active barcan dunes. In the central section of their distribution, the
oldest four ridges are continuous, but as they near Punta Parinas, the
first two disappear and the remaining two become a series of isolated
mounds. During this first episode of Holocene beach ridge develop-
ment, Punta Balcones and Punta Parinas were islands off the coast and
only during the emergence of the second group of ridges did these two
islands become landlocked.

These nine beaches were formed by the carrying of sediments from
the mouth of the Chira River, northward with the prevailing direction
of the current which were then implanted by wave action shoreward.
The rate of discharge of the Chira is second only to the Santa and the
Tumbes rivers and the sediment grains are uniform in size and have
the same constituents as the Chira River alluvial deposits (Chigne,
1975). The beach ridges were then raised by tectonic uplift and as a
consequence, a stranded beach system of 2.7 km wide resulted
(Fig. 2).

The archaeological sequence for the Piura-Chira Region has been
detailed elsewhere and the cultural associations of each ridge will be
briefly discussed but not elaborated upon (Tanning, 1963; Richardson,
1973, 1978). The following cultural sequence, based on forty radi-
ocarbon dates, will be used in this presentation: Amotape (11,500-

270

Annals of Carnegie Museum

VOL. 52

8000 B.P.); Siches-Estero (8000-5000 B.P.); Honda (5000-3500 B.P.);
Paita (3500-2500 B.P.), Sechura (2500-1200 B.P.); and Piura (1200-
1532 B.P.).

The shell capping of the beach ridges is the product of both natural
and human agencies. On all ridges there are fireplace units and small
middens of heated and discarded shells. On ridges 2-7, sherds are
present, and on ridges 8 and 9 lithic tools prevail. Many of these ceramic
fragments represent the breakage of whole vessles, due to wind erosion.
All of the complete vessels recovered were placed mouth downward
and buried, presumably for later reuse, and these are continually being
uncovered by the wind and rapidly eroded. Except for a market basket
handled Sechura pot and a badly eroded section of a Piura stirrup
spout, all of the pottery represents utilitarian ware; water storage jars
and a small percentage of wide mouthed cooking vessels. All of the
utilitarian ceramics have been tied into the ceramic assemblages of
each phase by comparison with nearby village sites and encampments.

The present population in Negritos and a small settlement on the
present beach dunes at the southern edge of the ridges, still exploit the
shellfish resources. Offshore fishing is also a major economic mainstay
of San Pedro, the small fishing village outside of Negritos (Sabella,
1974). Sabella (personal communication) has made the following com-
ments on the shellfish utilization by the Negritos population:

The main mollusk gathered is the concha blanca (Tivela hains) and
when they are in abundance, one man can fill an entire burlap sack
in 45 minutes. Concha blanca is gathered at low tide, gatherers move
into the shallows, first feeling for the shells with their feet. Once
located, they reach down and dig them out by hand, often pulling
several out at a time. The shells are relatively close to the surface,
in two or three inches of sand . . . these small clams are the preferred
bait for almost all types of fish, offshore as well as inshore species.
The second type of shell is very small and is referred to as rique
{Donax peruvianis). They are gathered with a ‘‘cafan” (a simple frame
scoop with a net bag attached, much the same as the “muy muy” or
sand crab scoop). The “cafan” is raked through the sand at low tide
several times, and then the bag is agitated to remove the sand, leaving
only the rique shells.

The beach ridges have been in continuous use since circa 5000 B.P.,
(Fig. 3). Prior to the beach ridge formation, the Lobitos Tablazo was
occupied by the preceramic Siches populations exploiting the mangrove
mollusks in the protected lagoon on the east side of the tablazo. The
small Siches site below Petro Peru marker 280 on the tablazo should
date between 8000-7000 B.P., because a Siches site (PV9-31) on the

i983

Richardson— Peruvian Maritime Economies

271

Fig. 2.— Air photo of the northern set of the Chira beach ridges. Photo courtsey of
Sevicio Aerofotografico Nacional del Peru, negative number 1662-D-l, 1946.

Talara Tablazo near the mouth of the Chira River is dated to 7485 ±
120 (SR 14 16) B;P.

With the emergence of the first Holocene shore line, the oldest ridge
was utilized by preceramic populations after the disappearance of man-
grove vegetation in the Talara region (Richardson, 1973, 1978). The
dates on ridge 9 are 3985 ± 80 (SI-1456) B.P., 4485 ± 90 (SI-1450)
B.P., and 4255 ± 65 (SI- 1420) B.P., which places the occupation of
this earliest ridge in the preceramic Honda phase. The base camp sites

272

Annals of Carnegie Museum

VOL. 52

I

Fig. 3. — The Chira beach ridges and associated radiocarbon dates.

i983

Richardson— Peruvian Maritime Economies

273

(PV7-16 and PV7-19) are situated on the Mancora Tablazo, north of
Quebrada Parinas and although no major Honda sites have been dis-
covered in the Chira Valley, they were certainly present and may have
been subsequently destroyed by later intensive irrigation farming.

Of the next three oldest ridges, ridge 6 is dehnitely correlated with
late Paita ceramics, 2685 ± 105 (SI- 1422) B.P. and on ridge 7 and 8
dates of 3500 ± 160 (GX1565) B.P. and 3490 ± 80 (SI-1421) B.P. are
fully acceptable for early Paita. Two dates, 3610 ± 145 (GXl 136) B.P.
and 3390 ± 125 (GXl 003) B.P. from the enormous Paita (PV8-7) site
at the south end of the ridges supports this interpretation. The hearth,
on ridge 8, from which the charcoal sample was secured, contained
two sherds of the Paita period, probably representing the reuse of this
ridge by Paita inhabitants of ridge 7. It is probable that ridge 8 may
also prove to be preceramic in its occupation.

The dates on the four oldest ridges, of course, refer to the period of
occupation and not to the date of the building or emergence of the
former beaches. The temporal formation for the hrst set of four ridges
spans over 3000 years.

Between this group of older ridges there is a 175 m wide separation
before encountering the youngest group (Fig. 4). The five youngest
ridges, including the present standline, begin their ceramic history dur-
ing the Sechura period. Ridge 5, first of this group, associated with
Sechura ceramics dates to 1955 ± 100 (SI- 1423) B.P., whereas ridge
4 dates to 1550 ± 110 (GX1556) B.P. and ridge 3 at 1405 ± 75 (SI-
1424A) and 1305 ± 100 (SI-1424B) B.P. It is interesting to note that
the Sechura ridges were as intensively occupied as the Piura ridges, but
that few village sites are known for the region versus the hundreds of
sites known for the Piura period. The Sechura sites on the Chira River
have most probably been destoryed through river channel changes and
modern agriculture in the floodplain zone. Ridge 2 dates to the Piura
phase at 805 ± 60 (SI- 1457) B.P. A circa 2000 year period is indicated
for the formation of this younger beach series. The width of these ridges
as compared to the older set are between 125 and 225 m wider, which
can be interpreted as the consequence of the Chira sediment load
spreading northward in conjunction with the development of dune
formations, not present on the older 4 ridges (Fig. 4).

The point to be stressed here is that the Chira beach ridges began
forming only when modern sea level was attained at circa 5000 B.P.
Certainly, beach lines were formed prior to 5000 B.P., but these were
either submerged as the ocean rose or were destroyed by wave action
as the sea encroached landward.

The beach ridges south of the mouth of the Chira River at Colan
are composed of pebbles which were distributed from a small canyon

274

Annals of Carnegie Museum

VOL. 52

Fig. 4. — Cross section of the Chira beach ridges at survey marker 280. Adapted from
Chigne (1975).

1983

Richardson— Peruvian Maritime Economies

275

cutting into the Talara Tablazo. The pebbles from the Talara formation
were eroded from the Canyon into the ocean at the southern edge of
the present Holocene deposits at Colan and distributed northward by
the current. The only mechanism of massive erosion in this desert
region are the El Ninos and there is a distinct possibility that the Colan
ridges were formed by flooding as a result of catastrophic El Ninos,
yet to be dated.

Conclusions

Our knowledge of the impact that natural forces had upon prehistoric
Peruvian cultural development is practically nil and until extensive
oceanographic, geological, and paleoclimatological work is done in the
Central Andes, the explanatory value of reconstructing Peruvian cul-
tural development, strictly from the archaeological record, may prove
misleading.

The date of 5000 B.P. is crucial to our understanding of the rise of
complex maritime societies on the coast of Peru, for it is at this time
that modern climate patterns had been established and modern coast-
lines formed. It was at this point in Peruvian cultural development
that complex maritime adapted societies were established. It is now
time for landlubbers to adapt themselves to the ocean and to interpret
maritime origins from a seaward position rather than to continue to
explain the establishment of maritime societies on the Peruvian coast,
strictly from a landward position.

Acknowledgments

I wish to express my thanks to Robert L. Stuckenrath of the Radiation Biology Lab-
oratory of the Smithsonian Institution for running the radiocarbon dates and to Michael
Moseley, for stimulating discussions on maritime origins. The interpretations and con-
clusions reached in this article are the sole responsibility of the author. The research
discussed herein was supported by the National Science Foundation, the University of
Pittsburgh Center for Latin American Studies, and Carnegie Museum of Natural History.
Figures were drawn by Frank Adkins.

Literature Cited

Bosworth, T. O. 1922. Geology of the Tertiary and Quaternary periods in the north-
west part of Peru. MacMillan and Company, London, 346 pp.

Chigne Campos, N. 1975. Movimiento de arenas y antiguas lineas de costa en el
noroeste Peruano. Unpublished B.A. thesis, Univ. Nacional de San Marcos, Lima,
77 pp.

Campbell, K. E. 1982. Late Pleistocene events along the coastal plain of northwestern
Peru. Pp. 423-440, in Biological diversification in the tropics (G. Prance, ed.),
Columbia Univ. Press, New York, 289 pp.

Cardenas Martin, M. 1978. Columna cronologica para el desierto de Sechura. In-
stituto Riva— Aquero Seminario de Arqueologia, Pontificia Univ. Catholica del
Peru, Lima 2:1-63.

276

Annals of Carnegie Museum

VOL. 52

Chostowski, M. n.d. The coastal morphology between Farinas and the Chira River.

12 pp.

DeVries, T. 1979. Nekton remains, diatoms, and Holocene upwelling off Peru. Un-
published M.A. thesis, Oregon State Univ., 67 pp.

Canning, E. P. 1963. A ceramic sequence for the Piura and Chira coast, north Peru.
Univ. California Publ. Anthropology, 46:135-284.

Lemon, R. R. H., and C. S. Churcher. 1961. Pleistocene geology and paleontology
of the Talara region, northwest Peru. Amer. Sci., 259:410-429.

Richards, H. 1962. Studies on the marine Pleistocene. Trans, of the Amer. Phil. Soc.,
52:1-141.

Richardson, J. B. III. 1973. The preceramic sequence and the Pleistocene and post-
Pleistocene climate of northwest Peru. Pp. 199-212, in Variation in anthropology
(D. Lathrap and J. Douglas, ed.), Univ. Illinois Press, Urbana, 111 pp.

. 1978. Early man on the Peruvian north coast, early maritime exploitation and

the Pleistocene and Holocene environment. Pp. 274-289, in Early man in America
from a circum-Pacific perspective (A Bryan, ed.), Univ. Alberta Press, Edmonton,
327 pp.

. 1981. Modeling the development of sedentary maritime economies on the

coast of Peru. Ann. Carnegie Mus., 50:139-150.

Rollins, H. B., D. Sandweiss, and J. B. Richardson III. 1981. A thermally-anomalous
assemblage (TAMA) from archaeological sites along coastal Peru: paleoecological
implications and biostratigraphical utility. Geol. Soc. Amer., 13:545.

Rollins, H. B., J. B. Richardson III, and D. Sandweiss. n.d. New evidence and
implications of a major change in the structure of the east Pacific water mass about
5000 B.P.

Sabella, j. 1974. The fisherman of Caleta San Pablo. Latin American Studies Program
Dissert. Ser., Cornell Univ., 57:1-342.

Sandweiss, D., H. B. Rollins, and J. B. Richardson III. 1983. Landscape alteration
and prehistoric human occupation on the north coast of Peru. Ann. Carnegie Mus.,
52:277-298.

Woodman, R., and M. Polia. 1974. Evidencia arqueologica del levantamiento con-
tinental al norte del Peru en los ultimos 4 mil anos. Bol. Soc. Geografica de Lima,
43:63-66.

7. 7 J

ISSN 0097-4463

Ai :nals

0/ CARNEGIE MLJSEUM

CARNEGIE MUSEUM OF NATURAL HISTORY

4400 FORBES AVENUE * PITTSBURGH, PENNSYLVANIA 15213

VOLUME 52 16 SEPTEMBER 1983 ARTICLE 12

LANDSCAPE ALTERATION AND PREHISTORIC HUMAN
OCCUPATION ON THE NORTH COAST OF PERU

Daniel H. Sandweiss^

Harold B. Rollins^

Research Associate, Section of Invertebrate Fossils

James B. Richardson III

Chief Curator, Section of Man

Abstract

A series of uplifted bays and former shorelines in the Chimbote region are discussed
in relation to tectonic uplift during the Peruvian preceramic period. The molluscan fauna
from a series of these preceramic sites are warm-water species, which the authors interpret
as evidence for a warm-water current prior to 5000 B.P. This indicates that the east
Pacific Ocean currents were markedly different from the modern current patterns. This
date has a major implications for the pre-5000 B.P. climate of the north coast of Peru.

Introduction

Interdisciplinary studies have long been recognized as essential in
understanding mankind’s prehistory. Archaeologists have increasingly
relied on the integration of many different branches of knowledge in
the attempt to interpret the past. In this article, the techniques and
contributions of archaeology, geomorphology, and molluscan paleo-
ecology have helped us to delineate the relationships between man and
a rapidly shifting physical environment during the preceramic epoch
on the north coast of Peru.

‘ Address: Department of Anthropology, Cornell University, Ithaca, NY 14835.

^ Address; Department of Geology and Planetary Sciences, University of Pittsburgh,
Pittsburgh, PA 15260.

Submitted 25 February 1983.

277

278

Annals of Carnegie Museum

VOL. 52

Geologic Setting

The Pacific Coast of Peru is tectonically active due to subduction of
the Nazca plate beneath the South American plate. The distribution
of earthquake hypocenters and volcanic activity suggest a large scale
segmentation into subunits with varying amounts of inclination be-
neath the South American plate (Barazangi and Isacks, 1976). Beneath
central and northern Peru, the Nazca plate is shallow and flattened and
is associated with a high level of shallow (50 km deep) seismic activity
but very little volcanic activity. In contrast, the Nazca plate beneath
southern Peru and northern Chile is steeply inclined with concomitant
deeper-focus seismic activity and increased volcanism. The boundary
between the two segments may represent a tear in the Nazca plate. The
shallow seismic zone beneath central and northern Peru may be the
most active of the world’s subduction zones (Barazangi and Isacks,
1976).

Human occupation of the Peruvian coast during the Holocene has
been accompanied by episodically intensive tectonism including oc-
casional cataclysmic earthquake activity (Moseley et al., 1981; Ortloff
et al., 1982; Sandweiss et al., 1981) and in some areas, volcanism. The
configuration of the subducting Nazca plate has not been geologically
stable. Although the region of central and northern Peru has not been
active volcanically during man’s habitation of the area, volcanic rocks
of Miocene and Pliocene age are common.

Tectonism, Molluscs, and Archaeology

Tectonism along the western margin of Peru has highly modified
coastal morphology, profoundly affecting the history of human occu-
pation in this region by altering natural and artificial environments
and the essential resources, which these environments contain. Many
changes visible in the archaeological record can be explained by ref-
erence to these environmental alterations. This paper details some
preliminary investigations into the relationships between landscape
alteration and human occupation on the north coast of Peru.

In order to understand changes in the resource base and settlement
patterns of prehistoric populations on the Peruvian coast, it is necessary
to temporally correlate tectonic episodes with archaeological sites. Ma-
rine molluscs, which are often very habitat-specific and have long been
exploited by coastal populations, provide a valuable index of changes
in coastal geomorphology and ecology and therefore provide powerful
chronostratigraphic and paleoecologic tools.

1983

Sandweiss et al. — Human Occupation on Peruvian Coast

279

Fig. L — Project location. The Pam.pas las Salinas and Salinas de Chao sites are indicated
by X’s on the inset map.

The results of our investigation of preceramic sites on the northwest
coast of Peru support a tectonic model for the area which involves
relatively small, but distinct, blocks, possibly bounded by major fault
systems trending approximately east-west (perpendicular to the coast).
The rivers along the Peruvian coast also run east-west and may be
controlled by these faults. Each block appears to have had a different
tectonic history, decipherable by the study of molluscs in prehistoric
midden deposits and in living position on uplifted beaches and bay
floors, in conjunction with the study of relict geomorphic features vis-
ible in the field and on aerial photographs. The sparsely vegetated arid
region of coastal Peru is an ideal setting for preservation and obser-
vation of such features.

This study focuses on two elevated relict bays in the Chao and Santa
valleys (Fig. 1) and draws comparisons with other uplifted bays at
Ventanilla on the central coast (Moseley, 1975) and at Otuma on the

280

Annals of Carnegie Museum

VOL. 52

south coast (Craig and Psuty, 1971) and with the still-active bay of
Huaynuna, between the Casma and Nepeha valleys to the south of the
Santa River. The relict bays of Santa and Chao are only 1 5 km apart.
Although both have evidence of preceramic occupation prior to uplift,
the ancient strandline collectors of Chao gathered a diversity of cool-
water molluscs indigenous to the present-day Peruvian coast, whereas
the Santa strandlopers exploited an anomalous molluscan fauna.

The very different setting of Huaynuna Bay, also with associated
preceramic middens, suggests the presence of a tectonic province
boundary between Huaynuna and the Santa-Chao area to the north.

Landscape Alterations and Human Occupation

Pampa las Salinas, Santa Valley

Situated to the north of the modern Santa River mouth, the present-
day Pampa las Salinas area is a barren desert zone characterized by
windswept beach ridges, salt flats, and rocky foothills (Fig. 2). The area
has no natural source of surface water, and human habitation is re-
stricted to a few dwellings in the irrigated southern portion of the Pampa
along with several chicken ranches located further to the north, on top
of relict beach ridges. The beach ridges provide access roads to the
shore and a source of beach cobbles for construction. The salt flats are
currently worked with heavy machinery.

A few thousand years ago, this region was geomorphically, and pos-
sibly climatically, quite different. The shoreline at that time was situ-
ated about 5 km to the east of its present position and formed a long,
curving bay at the base of the foothills (Fig. 2). Fronted by an ancient
sea cliff 5 to 10 m high, this relict bay represents the highest shoreline
stand in the region. Sedimentary deposits on top of the sea cliff consist
of culturally accumulated shell middens on top of colluvium naturally
derived from the adjacent foothills. Below the cliff, the ancient in situ
beach deposits are preserved as a white band of largely broken and
size-sorted shell material, visible even on aerial photographs.

Preceramic shoreline collectors deposited refuse in middens along
the top of the sea cliff and maintained a base camp on lower ground
at the southern end of the bay. The largest of the cliff-top middens, the
Ostra site, is situated on and around a rocky promontory overlooking
the ancient beach.

Excavation of a small test pit revealed 60 to 70 cm of unstratified
shell debris, some fish remains, and charcoal. No artifacts were found.
Two C-14 assays on shell material from the Ostra midden yielded
uncorrected dates of 5400 ± 60 years B.P. (SI-4954) for the bottom
of the deposit and 5160 ± 60 years B.P. (SI-4955) for the middle of
the deposit. The molluscan assemblage consisted entirely of warm-
water, species, most of which are now only found 500 km or more to

1983

Sandweiss et al. — Human Occupation on Peruvian Coast

281

Fig. 2. — The Pampas las Salinas preceramic sites. The Ostra base camp and collecting
station and the borrow pit are denoted with X’s. The contour interval is 100 m.

282

Annals of Carnegie Museum

VOL. 52

Table \.—Molluscan species recovered from the Ostra Site, Pampa las Salinas.

Species

Modern distribution

Argopecten circularis (scallop)

north of Paita, Peru

Ostrea chilensis (oyster)

entire coast, mostly
north of Paita

Anomia sp. cf A. peruviana

north of Paita

Trachycardium procerum (Clam)

entire coast

Chione (Lirophora) discrepans (Clam)

north of Islay

Protothaca (Colonche) ecuadoriana (clam)

north of Tumbes

Anadara bifrons (clam)

north of Paita

Cerithium (Thericum) stercusmuscarum (snail)

northern Peru

Cerithidea mazatlanica (snail)

north of Peru

Nassarius (Arcularia) tiarula (snail)

north of Peru

the north of the Santa Valley (see Table 1). Most abundant were oyster
and scallop shells, neither of which had previously been reported from
any other archaeological site along the Humboldt-washed Peruvian
coast. Subsequent to our work at Pampas las Salinas Pozorski and
Pozorski (n.d.) have found an undated assemblage of warm-water mol-
luscs at the preceramic Almejas site near Huaynuna bay.

The cockle Trachycardium procerum was the only mollusc species
at the Ostra site that has also been found at other sites within the cold-
water Peruvian province. T. procerum is quite temperature tolerant
and was noted by Olsson (1961:247-248) to occur in both the warm-
water Panamic and cool-water Peruvian provinces. Over its current
geographic range, it displays the often-noted trend among molluscan
species of increased size in colder waters. Olsson (1961) stated that
“Peruvian specimens are as a rule larger than those from more northerly
stations.” The Ostra specimens are smaller than those from other Pe-
ruvian sites, in accordance with other evidence of warm-water con-
ditions.

All of the mollusc species found in the Ostra site were also collected
from the relict beach area directly below the site. Frequency distri-
butions were, as expected, different in the two assemblages, reflecting
human selection for economically more important (that is, larger)
species. Some of the species were in living orientation in the relict
beach area. Oysters, for example, were found cemented to isolated,
often partially buried rocks. Excavation of shallow trenches at right
angles to the paleo-strandline displayed typical flaser interbedding of
intertidal and shoreface sands.

This undisturbed preservation of the beach facies mosaic, with no
overlying sedimentary sequence, attests to either very rapid tectonic
uplift of the area or a sudden sea level drop. The evidence demands a

1983

Sandweiss et al. — Human Occupation on Peruvian Coast

283

rapidity of change that is difficult to attribute to non-tectonic causes.
Moreover, strandline retreat via sedimentary progradation would have
most likely buried or destroyed the integrity of the facies mosaic by
erosional reworking. The relative uplift of the Ostra site and concom-
itant westward migration of the strandline will hereafter be referred to
as the Salinas Event. We believe that the suddenness of the Salinas
Event is most compatible with a tectonic cause of cataclysmic nature.

The presence of a warm-water molluscan fauna at the Ostra site
invites further speculation (Rollins et al., 1981). There is a substantial
literature dealing with thermally anomalous molluscan assemblages,
herein referred to as TAMAs (see Zinsmeister, 1974, for a recent re-
view). In each described TAMA occurrence, the anomalous assemblage
consists of a very few species and very few specimens of any particular
species. With its moderately high diversity of Panamic (warm-water)
species (see Table 1), the Ostra occurrence is a unique TAMA. More-
over, it obviously persisted at the Ostra site long enough to become a
local resource base for a preceramic population. All previously de-
scribed TAMAs, such as those of the early Pleistocene of California,
can be attributed to individual years when warm (or cold) currents
varied in intensities and positions. This permitted pelagic larvae, es-
pecially the planktotrophic forms with extended planktonic stages, to
temporarily extend their ranges. All such TAMAs would be short-lived,
incapable of becoming a significant food resource of the kind indicated
by the Ostra remains. The Ostra occurrence requires an explanation
which would permit the displaced fauna to spawn and increase pop-
ulation size sufficiently to allow profitable harvesting. All other de-
scribed TAMAs involved species being transported into water where
normal temperatures prohibited reproduction. The Ostra site and as-
sociated relict beach indicate the persistence of warmer (reproductively
compatible) water conditions for at least several years— multiple age-
classes of the transitional, warm-water species are preserved in the
relict beach assemblage.

The coast of Peru certainly provides ample opportunity for current
displacement of molluscan larvae. The sporadic El Nino current brings
warm water down the Peruvian coast from the Gulf of Guayaquil area.
Maximum El Nino events can reach as far south as the Paracas pen-
insula, although such events occur on the average only several times
per century. A live Spondylus mollusc was reported from Callao Harbor
on the central coast following the strong El Nino occurrence of 1925
(V. Valdiviezo, personal communication). Spondylus is a Panamic
province mollusc normally limited to the warm waters north of Paita.

One explanation for the Ostra TAMA occurrence would involve
closely-spaced, multiple El Nino events bringing warm-water mollus-
can larvae into the Ostra Bay, which might have been shallow enough

284

Annals of Carnegie Museum

VOL. 52

to be heated by the sun. This explanation, however, requires that El
Nino currents operated in the past in ways signihcantly different from
what is indicated by their known history.

We believe that the Ostra TAMA occurrence is best explained as the
result of a radically different organization of currents along the Peruvian
shore prior to about 5000 B.P. According to our reconstruction, a
southward-flowing Equatorial Countercurrent system formerly washed
the coast of north-central Peru and supported a warm-water, Panamic-
Peruvian molluscan fauna. Only after 5000 B.P. did the boundary
between the Humboldt-dominated, cool-water current system and the
warm-water system shift 400 km north to its present location near
Paita. Alterations in the Quaternary distributions of ocean-floor phos-
phorite deposits and land-based guano-derived phosphate deposits sup-
port the TAMA evidence of permanent warm- water conditions along
the early Holocene Peruvian shore (Burnett, 1980). The arguments for
the postulated oceanic and climatic alterations are detailed in Rollins
et al. (n.d.)

Paleoenvironmental relationships of the preceramic Ostra base camp
site are more difficult to determine, due to later wall and road con-
struction, cultivation, and extensive looting. Surface observations in-
dicate that the base camp middens contained a molluscan fauna similar
to that at the Ostra site, as well as fish and sea mammal remains and
a crude lithic industry composed of cores, hammerstones, and unifacial
scrapers made from beach cobbles. Mortars are also present and may
have been used to process fish and molluscan resources.

The increased diversity of resource species coupled with the presence
of artifacts suggest that this site was actually a base camp for the
preceramic people who maintained a string of collecting stations such
as the Ostra site along the bay shore to the north. The base camp was
thus situated on the bay as near as possible to the mouth of the Santa
River. This location would have maximized access to both bay shore
and riverine environments and their respective resources.

The Salinas Event displaced the shoreline several kilometers to the
west, probably causing the destruction of the warm-water mollusc beds
and the abandonment of both the Ostra site and the Ostra base camp.
There is no indication at either site of a shift in harvested species from
warm water to cold water shellfish. This fact supports the argument
that the sites were abruptly abandoned prior to the shift in coastal
currents, probably following and as a result of the Salinas Event.

The Salinas Event also triggered two related geomorphic processes
in the Pampa las Salinas area— the progressive southward migration
of the Santa River mouth and the westward progradation of the coast
due to sequential beach ridge formation. Prior to the uplift, the Santa
River may have entered the ocean through the ancient bay itself (Mose-

1983

Sandweiss et al. — Human Occupation on Peruvian Coast

285

ley, personal communication). Evidence for this course is slight at
present, but the suggestion deserves serious consideration. The presence
of an active river mouth in the bay would affect interpretations of Ostra
settlement patterns.

Immediately after the uplift, the Santa entered the ocean from behind
a hill on the southern end of the Pampa. This mouth (Mouth 1) may
have been the pre-uplift course as well. In any event. Mouth 1 was the
basal focus of the first series of beach ridges visible in aerial photographs
of the Pampa las Salinas. Later, the river mouth shifted several hundred
meters south to exit behind another hill (Mouth 2), generating a second
series of beach ridges. Mouth 1 is cut across the base, presumably by
the river when it flowed from Mouth 2. This fact, along with the
implications of the east to west (early to late) temporal gradient of the
beach ridges, indicates that Mouth 1 predates Mouth 2.

More recently, the river has continued it southward migration to its
present position 3 km south of Mouth 1 . The lack of beach ridge
formation in the area between Mouth 2 and the present mouth suggests
that this last shift has occurred fairly recently. A few small, incipient
ridges have formed to the north of the present mouth. Decreased flow
due to bleeding of the Santa for irrigation may have affected the river’s
capacity to carry material for beach ridge consturction; progressive
stabilization of the drainage system following the Salinas Event also
might have caused a systematic reduction in the amount of material
available for transport and incorporation in new ridges.

The southward migration of the Santa River following the Salinas
Event can be explained by several models. The Salinas Event may have
caused (or emphasized a pre-existing) south-dipping tilt of the coastal
plain in the Santa Valley region. Alternately, if the Santa River runs
along a major fault line perpendicular to the coast, uplift of the northern
block relative to the southern block could have caused the river to
move southward off the higher northern block. Further field work is
necessary to test these various hypotheses.

Beach ridge formation is closely related to the actions of the river.
The ridges all originate at river mouths and flow northward in the
direction of the prevailing, wave driven longshore current. At least nine
ridges run the length of the coast on the westward margin of Pampa
las Salinas. Each ridge is composed of marine sand and river cobbles.
Active marine involvement in the formation process is shown by the
presence of several small, intertidal gastropods {Prisogaster niger and
Tegula atra) still associated with their operculah Each ridge, as it was

‘ Opercula are small discs used by certain gastropods to seal off the entrance to their
shells. Because they are attached only by soft parts and become automatically disartic-
ulated after death, the proximity of the opercula to the shells is a strong indication that
the gastropods have not been transported since death and that the deposit is therefore
primary.

286

Annals of Carnegie Museum

VOL. 52

constructed, must have been right on the beach front. The construction
of a new ridge would deactivate the preceding one, creating a time
transgressive, horizontally stratified sequence moving from east (ear-
liest) to west (latest). Shell material from a borrow pit in one of the
earliest ridges yielded an uncorrected C-14 date of 4235 ±115 years
B.P. (SI-4957). This date falls nearly one thousand years and many
standard deviations later than the most recent date for the uplifted
Ostra site, confirming the geomorphic observation that the ridge for-
mation began subsequent to the Salinas Event.

Two processes may be involved in the sequential formation of the
Santa beach ridges. Grolier et al. (1974) feel that the multiple ridges
resulted from episodic seismic events, with each ridge representing a
new, minor uplift. The other possible process is lateral accretion, the
way in which beach ridges prograde seaward in tectonically stable areas
(see for example. King, 1973). The beach ridge borrow pit mentioned
above revealed a diagnostic sedimentary sequence. A thick layer of
poorly sorted, relatively fine material with a few poorly rounded large
clasts was overlain by a well sorted layer of well rounded cobbles. The
following scenario would account for this sedimentary sequence:

1 . A strong El Nino event accompanied by abnormal precipitation
washed large quantities of loose material into the river from the nor-
mally inactive, unvegetated, arid region of the lower Santa drainage
basin, while simultaneously increasing the competence of the river.
Imbalances in the drainage system caused by tectonic episodes such as
the Salinas Event would have increased still further the amount of
material available for transport. The normal lack of rain means that
the drainage system only adjusts itself much more rapidly to landscape
alterations of tectonic origin during the rare El Nino events accom-
panied by heavy rains than at other times.

2. The rapid efflux of material out of the river mouth caused rapid
progradation of the coast to the north of the river as the material was
transported and deposited by the longshore current. This material was
essentially unmodified and poorly sorted, similar to the lower layer
found in the beach ridge.

3. After the El Nino rains ceased and the coastal system was no
longer swamped with sediment, wave action again dominated coastal
processes. The result was that fine sediment was removed and larger
clasts were rounded and thrown up on the shore as a lag-type deposit.
The upper layer of the beach ridge represents this kind of deposit. After
each ridge reached a threshold height, material no longer reached the
top and a new ridge formed on the seaward side with the next El Nino.

Both episodic, minor uplift and lateral accretion relating to major
El Nino events may have been involved in the observed sequential
formation of beach ridges in the Pampa las Salinas area. Further field

1983

Sandweiss et al. — Human Occupation on Peruvian Coast

287

testing will be required in order to evaluate the contribution of each
of these processes. If the El Nino hypothesis proves valid, then the
Santa beach ridges encode data on the history of major El Nino oc-
currences over the last 4000 years. Each major El Nino could be dated
by C-14 analysis of marine gastropods incorporated in the diagnostic,
El Nino-derived sedimentary sequence. These studies would supply
information on intensity and periodicity of strong El Nino events. Such
information is of considerable consequence in understanding the de-
velopment of coastal Peruvian societies (Sandweiss, n.d.b.).

The Pampa las Salinas beach ridges have almost no evidence of pre-
modern human use. The only abandoned structures on the ridges are
a few slightly curved lines of stones opening away from the prevailing
southwesterly winds; these lines may have been the foundations of
windbreaks. They have no associated artifacts and could as easily be
modern as ancient. The lack of sites in the Pampa las Salinas following
the Salinas Event suggests a dwindling resource base related to the large
magnitude uplift and perhaps to the postulated shift in coastal current
regimes.

Salinas de Chao, Chao Valley

The other raised bay under consideration lies 1 5 km north of Pampa
las Salinas, at Salinas de chao. Prior to uplift, this relict feature was a
deep, arcuate bay fronted by a 5 to 10 m high sea cliff and anchored
at the south end by a large, rocky promontory. While active, the Salinas
de Chao Bay must have resembled the modern situation at Guahape,
40 km north in the Viru Valley. We refer to the uplift which stranded
the Salinas de Chao as the Chao Event^.

Today, the outlines of the relict bay are clearly visible in aerial
photographs. Massive dunes stretch across the former floor, which now
contains commercially exploited salt flats. A faint set of beach ridges
intervenes between the flats and the modern shore. As in the Pampa
las Salinas, the Salinas de Chao ridges must have formed after the uplift
and are probably an actual extension of the southern (Santa) ridges.
The Santa River is the nearest source of raw material upcurrent from
the Salinas de Chao; the distance from this source accounts for the
diminished size and definition of the Chao ridges.

A number of archaeological sites are located in the Salinas de Chao
area. Most are preceramic. Based on surface collections of molluscs
and the postulated effects of the Chao Event on the natural molluscan
association, we proposed a relative chronology for four of these sites.

^ Thomas J. DeVries (personal communication) believes that there may have been two
episodes of uplift at Satinas de Chao, separated by decades or even several hundred
years. Again, further fieldwork will be necessary to evaluate this possibility.

288

Annals of Carnegie Museum

VOL. 52

Fig. 3. — Location of the Salinas de Chao preceramic sites. Contour interval is 100 m.

Radiocarbon dating of shell material from three of the sites has partially
conhrmed our analysis.

Called Los Morteros by Cardenas (1979) and referred to here as Site
B, the earliest site is a large mound on the edge of the former sea cliff
(Fig. 3). A rocky knoll may underlie the mound. Architectural remains
at Site B consist of several crude, round foundations. Two C-14 dates
on shell material from the surface of this site yielded uncorrected dates
of 4380 ± 75 years B.P. (SI-49598) and 4040 ± 75 years B.P. (SI-
49 5 9 A), while a charcoal sample was assayed at 4010 ± 85 years B.P.
Cardenas (1979:28) reports two dates for the site, 4560 ± 60 and 4660
± 60 years B.P. The earlier date is on charcoal from a depth of 1.20
m and the later date is on charcoal at a depth of .40 m in the same pit
(Cardenas, 1977-1978:26-27).

Molluscs found on Site B cover a broad spectrum of cool-water
species from both sandy and rocky habitats (see Table 2). Many mol-
luscs are habitat-specihc, and sand-dwelling species are usually limited
to sandy beaches, whereas rock-dwelling species live on rocky pro-
montories and similar zones. Finer habitat distinctions than the basic
sand/rock dichotomy are possible, but are unnecessary for the present
analysis. Food is the only indicated use of molluscs in the Salinas de

1983

Sandweiss et al. — Human Occupation on Peruvian Coast

289

Table 2.—MoHuscan species recovered from Site B, Salinas de Chao.

Species

Modern distribution

Argopecten purpuratus (scallop)

south of Paita, Peru

Donax peruvianas (Clam)

Ecuador to Chile, rare
north of Paita

Mesodesma donacium (clam)

south of Paita

Barbatia (Acar) pusilla (clam)

south of Paita

Trachycardium procerum (clam)

entire coast

Chormytilus chorus (mussel)

south of Paita

Concholepas concholepas (snail)

south of Callao

Sinum cymba (snail)

Ecuador to Chile

Thais (Stramonita) haemastoma (snail)

Peruvian-Panamic province

Thais {Stramonita) chocolate (snail)

entire coast

Mitra (Atrimitra) orientalis (snail)

south of Paita

Fissurella crassa (limpet)

south of Galapagos

Scurria parasitica (snail)

south of Paita

Polinices (Polinices) intermeratus (snail)

Baja California to Peru
(14H4'S)

Crepidula cf. C. onyx (snail)

southern California to Chile

The following additional species were identified at Site B, Salinas de Chao by Thomas
J. DeVries (personal communication):

Xanthochorus buxea (snail)

Perumytilus purpuratus (mussel)

Aulacomya ater (mussel)

Diloma nigerrima (snail)

Tegula atra (snail)

Polinices uber (snail)

Amphineurid spp. (chitons)

All of these species are Peruvian.

Chao sites, so it is safe to assume that the shellfish were collected live
from their respective habitats (rather than as beach detritus) and there-
fore indicate the range of microenvironments exploited at the time of
site occupation.

According to our molluscan model, Site B was occupied prior to
uplift, while the bay was still active. The site’s location gave it direct
access to the sandy littoral zone in both directions, whereas a rocky
shore environment lay within easy walking distance to the southwest.
The modern fishing hamlet of Puerto Moorin on Guanape Bay in the
Viru Valley is similarly situated; fishermen routinely walk to the rocky
headlands to collect molluscs and to fish (Sandweiss, n.d.a.).

Various factors may have influenced the choice of site location on
the bay shore some distance from the interface between the rocky and
sandy littoral. The following criteria were probably significant:

1. The location of fresh water must have been important. Unfor-

290

Annals of Carnegie Museum

VOL. 52

tunately, we have no information on the location of water sources in
the Salinas de Chao.

2. The rocky littoral zone has a more limited spatial distribution
than the sandy shore, thus concentrating species available on or from
rocky headlands into a higher density biomass. Mosely (1975:14) has
written that ‘Tor early populations, one of the richest of all local eco-
nomic complexes” was the rocky shore. Sandy shore resources are more
spread out; sand-dwelling mollusc beds have a tendency to move from
place to place. This kind of shifting, extended distribution necessitates
a large collecting range. Site B may have been located to allow access
to the maximum possible amount of sandy littoral zone while still
keeping the rocky shore resources within practical exploitation range.

3. The rocky knoll suspected to underlie the site is the only such
feature along the shore of the relict bay. Considering the similar situ-
ation of the Ostra site, it seems likely that rocky knolls overlooking
the beach were the preferred settlement locations of preceramic strand-
line collectors. Advantages of a raised dwelling site along the coast
include a better view of the surroundings and approaches, protection
from waves and extra-high tides, and exposure to the sea breezes which
drive off mosquitos.

4. Site B is located on that section of the relict bay which lies closest
to the most obvious access route to the Chao Valley and to the interior
in general. This fact suggests that the site may have been occupied
seasonally by people moving down from more inland regions.

Site B occupation terminated with or before the stranding of the bay
by the Chao Event. No other site has as large a variety of molluscs,
either in terms of number of species or of exploited microenvironments;
destruction of molluscan resources by a tectonic event is implicated in
the decrease in diversity. The Chao Event therefore dates after about
4000 years ago, more than 1000 years after the Salinas Event. The
Pampa las Salinas displays no evidence of a major uplift following the
Salinas Event. Therefore, the time gap between the two named events
indicates that the two zones have distinct Holocene tectonic histories
despite their geographic proximity.

Following the malaco-chronologic model, the next site in the se-
quence is Site C, on the shore of the relict bay to the southwest of Site
B. The molluscan assemblage consists almost exclusively of two sand-
dwelling beach clams, the small Donax peruvianus and a larger bivalve.
Only a very few specimens of rock-dwelling shellfish {Fissurella sp. and
Conchelepas conchelepas) are present, despite the location of Site C
closer to the rocky headlands than Site B. Architectural remains consist
of crude, round foundations of stone, similar to the surface architecture
of Site B.

1983

Sandweiss et al. — Human Occupation on Peruvian Coast

291

The following scenario may explain the transition from Site B to Site
C, if the two sites really are sequential. First, the Chao Event uplifted
the bay, displacing the shoreline to the west. At the same time, the
rocky promontory at the southwest end of the bay was uplifted suffi-
ciently to strand the intertidal zone and kill the resident rock-dwelling
molluscs. The Site B occupants, perhaps after a short hiatus, moved
to the Site C location to be closer to the new shoreline and its con-
comitant marine resources. Destruction of the intertidal mollusc pop-
ulations accounts for the lack of rock-dwelling species in the site. Be-
cause rocky headlands on the Peruvian coast are discrete units separated
by large expanses of sandy beach, it is difficult for rock-dwelling mol-
luscs to recolonize a rocky zone immediately following destruction of
the resident population. The loss of variety in the sand-dwelling species
may have resulted from destruction of the bay bottom shellfish beds
and from differential speeds of recolonization — beach clams are fast
colonizers and would be the first species to appear in quantity following
a catastrophic uplift. Water source remains an unknown factor. The
decision to relocate on the now-stranded sea cliff fronting the relict bay
rather than the open bay bottom nearer to the shore may be explained
by the apparent preference for raised dwelling sites and the probable
instability of the immediately post-uplift bay floor.

The one radiocarbon date for Site C contradicts the details of the
above scenario. A sample of shell material from the surface of the site
yielded an uncorrected date of 2980 ± 115 years B.P. (SI-4958), a
thousand years after the dates for Site B. Either the date is in error
(possibly the dated shell was dropped by a later collector passing over
the site long after it was abandoned) or else our reconstruction must
be revised. Even if the date is correct. Site C still post-dates Site B and
the loss of variety in mollusc species can still be explained by tectonic
destruction of the local rocky habitat.

The third site in the sequence (Site D) is located on the ancient sea
cliff about 100 m to the north of Site B. The molluscan fauna at this
site are the same as in Site C, with the addition of another sand-dwelling
clam, Mesodesma donacium. Building remains at this site consist of
well constructed, rectangular stone foundations. We consider this site
to be later than Site C on account of the addition of M. donacium to
the molluscan assemblage and because the architecture differs from
that of Site B and Site C but is similar to that of the Las Salinas site
(Site A, see below). Comparison of the molluscan assemblages reveals
a temporal discontinuity between Site B and Site D, despite their spatial
proximity. If the sites had been contemporaneous, they would not have
had such a marked difference in the number and kind of exploited
molluscan species. No radiocarbon dates are available for Site D.

292

Annals of Carnegie Museum

VOL. 52

Called Las Salinas by Cardenas (1979), Site A is a large preceramic
center with complex architecture including rectangular rooms and pla-
zas, a sunken circular court, and platforms (Alva, 1978). Cardenas
(1979:28) has obtained 6 C-14 dates for this site; the dates range from
3570 ± 60 B.P. to 3300 ± 150 B.P. and average about 3400 years
B.P. These dates fall over 500 years after the Site B dates, at the very
end of the Preceramic period.

Site A lies about 1 km to the east of the raised bay and is built on
and around a rocky spur at the entrance to the Salinas de Chao. The
molluscan remains show a greater variety of species than either of the
other post-uplift sites, but fewer species are present here than in the
pre-uplift Site B. The increase over earlier post-uplift times reflects in
part a restablization of the local mollusc populations. The increase in
site size and complexity evident in the layout of the Las Salinas site
must reflect an increase in both the size and the organizational com-
plexity of the Salinas de Chao population. These changes would support
an increased exploitation range, which may account in part for the
greater diversity of molluscan species. Considering modern conditions,
it seems unlikely that the rocky promontory nearest to the Salinas de
Chao provided a suitable habitat for rock-dwelling shellfish at any time
following the uplift, so the presence of rocky shore molluscs in Site A
strongly supports the hypothesis of expanded exploitation range. Be-
cause Site D lies on the direct route from Site A to the shore, the fact
that Site D has the lesser variety of molluscs suggests that the two sites
are not contemporaneous. An alternate explanation is that the two sites
were occupied at the same time but that only Site A commanded the
necessary labor or territorial rights to exploit or acquire resources from
outside the immediate Salinas de Chao vicinity.

The size and complexity of the Las Salinas site and its location away
from the shore but strategically placed to control access to the Salinas
de Chao suggests the presence of some important resource in this area.
Not far from the site is a wall, now partly destroyed, which blocks the
entire access route from the interior. The most obvious candidate for
the controlled resource is salt from the stranded bay floor. Access to
salt was very important and strickly regulated among Andean groups
at the time of the Spanish conquest and probably for many years before
that time. If this interpretation is correct, it demands a post-uplift date
for Site A: the key resource, salt from the salt flats, was only available
after the Chao Event cast the bay floor up from the sea.

In summary. Site A is differentiated from the other three sites on
the basis of the molluscan assemblages, the relative size and complexity,
the location away from the shore of the relict bay, and the C-14 dates.
Site A is the largest site of any period in the Salinas de Chao region.
Later sites include an Early Intermediate shell midden near the modern

1983

Sandweiss et al. — Human Occupation on Peruvian Coast

293

shore containing great quantities of the sand-dwelling clam Mesodesma
donacium, but no visible architectural remains. This site is interpreted
as a specialized shellfish collecting (and drying?) station. Site A had a
late reuse by squatters who left some pots and scatters of the small
beach clam Donax peruvianus on top of several of the platforms. The
pottery is in the Casma Incised style and dates to the Late Intermediate
period (C. Daggett, 1983).

Huaynuna, Casma- Nepena Valleys

Fifty km south of the Santa River, the active bay of Huaynuna lies
between the Casma and Nepena Valleys. A preceramic midden is sit-
uated only a few meters above and behind the bay shore. The midden
contains abundant organic remains including the shells of cool-water
molluscs. Sand-dwelling species predominate, but rock-dwellers are
also present. The site in underlain by alluvium, beach sand, and an
indurated sandstone (beach rock) containing large scallop valves {Ar-
gopecten purpuratus). The scallop valves were deposited on a bay floor
and are C-14 dated at 3635 ± 80 years B.P. uncorrected (SI-4956).
Because the indurated sandstone lies near the present sea-level and
because the alluvial deposits on top of it are of non-marine origin,
Huaynuna Bay seems not to have been significantly uplifted during the
last 3500 years. In fact, none of the indicators of massive uplift seen
in the Santa-Chao region are present here — no raised beaches or sea
clifls, no stranded shellbeds, and no beach ridges are found at Huay-
nuna. The presence of a midden so close to the modern shore shows
that the coastal configuration has not changed significantly since the
site was occupied during preceramic times. The hills around Huaynuna
appear more rounded than those in the Santa-Chao region, indicating
a longer term of stability and erosion in the drainage system of the
former area (J. Donahue, personal communication).

Dating of the major Holocene tectonic events in the Santa and Chao
zones shows that the two areas differ in terms of the chronology of
seismic activity. Comparison of Holocene uplift features in the Santa
and Chao zones with the lack of such features around Huaynuna in-
dicates notably divergent tectonic histories, an order of magnitude
greater than the differences between Santa and Chao. This fact suggests
that the Huaynuna and Santa-Chao areas belong to different tectonic
blocks. The boundary between these provinces may be a major fault.
The Santa River may lie along such a boundary; the Canon del Pato,
a narrow gorge where the Santa comes through the Cordillera Negra,
is highly suggestive of a major fault. A smaller fault may separate the
Santa and Chao regions and may be indicated by the rocky spur which
shears off from the coastal range to form the headland south of Salinas
de Chao.

294

Annals of Carnegie Museum

VOL. 52

Other Uplifted Bays

Several other raised bays with associated preceramic middens have
been reported on the Peruvian coast. Pozorski and Pozorski (1979:
337-341) use settlement pattern data and subsistence remains to sug-
gest a date of around 0-200 A.D. for major tectonic activity at Salaverry
in the north coast Moche Valley. They go on to list a number of
segments of the Peruvian coast possibly uplifted during the late Ho-
locene. These segments range from the far north coast to the south
coast, and all have associated archaeological sites.

Moseley (1975) has discussed a number of sites on the shore of the
raised bay at Ventanilla, to the north of the Chillon Valley on the
central coast. He feels that the bay was gradually stranded by a “marine
regression” which “left the rocky habitat dry and stranded” (1975:23).
“When the waters receded, vast beds of clams were left stranded on
the sand flats where they remain to this day” (1975:25). This situation
recalls Pampa las Salinas and Salinas de Chao.

Moseley (1975:51) goes on to discuss two sites which lie within 2
km of each other but contain distinct molluscan assemblages. The
Camino site has exclusively sand-dwelling species, while at the Pampa
site, rock-dwellers predominate although sand-dwellers are also pres-
ent. Radiocarbon dates “suggest overlapping occupations.” Because
the two sites are close together and have similar access to both sandy
and rocky littoral resources, Moseley considers them to represent “the
earliest possible suggestions of differential resource accessibility,” per-
haps “the result of the imposition of jural rights through which the
residents of each settlement controlled access to resources in their
immediate vicinity.” As an alternative, he suggests that “the occupa-
tional overlap between Pampa and Camino implied by the radiocarbon
dates is more apparent than real, in which case differences in subsistence
patterns probably reflect factors other than jural rights.”

Moseley is probably correct that “factors other than jural rights” are
the cause of the observed differences in molluscan assemblages. A
sudden uplift such as occurred in the Santa and Chao areas would have
stranded the rocky habitat instantaneously, necessitating an immediate
shift to the exclusive exploitation of sand-dwelling species. The rapidity
of this change would explain the similarity in C-14 dates. According
to the uplift hypothesis, the two sites are sequential rather than over-
lapping, with Pampa preceding Camino. This situation parallels the
apparent sequence in the Salinas de Chao.

Craig and Psuty (1971) discuss a raised bay and associated preceramic
shell middens at Otuma on the south coast of Peru. Their report pre-
sages many of the concerns of the present study. At Otuma, 3 1 out of
32 mounds of scallop {Argopecten purpuratus) valves are

1983

Sandweiss et al. — Human Occupation on Peruvian Coast

295

“situated at the crest of a raised sea cliff that nearly encircles the
stranded Otuma embayment . . . stratigraphic analysis of midden
contents indicates the Otuma lagoon was stranded by a sudden uplift
that destroyed its shellfish beds. Abrupt abandonment of the locale
by the aboriginal inhabitants followed” (1971, 152).

Limited isotopic dating of shell carbonate gave five dates between
3100-3600 B.P. (Craig and Psuty, 1971:130). The situation at Otuma
is quite similar to that at the other raised bays, the closest parallel
being with the Pampa las Salinas.

Conclusions

Both the data and interpretations presented in this paper are prelim-
inary in nature and serve mainly to raise questions and suggest new
ways of approaching prehistoric phenomena in coastal regions of in-
tense tectonic activity. Molluscan analysis not only indicated a signif-
icant paleoclimatic anomaly, but also proved to be a useful tool for
studying the relationships between human occupations and coastal
landscape alterations. In the Santa and Chao areas, the presence of shell
middens some distance from the modern shoreline was the key to
recognizing the Holocene date of drastic changes in coastal morphology.

We have identified three major, interrelated processes of landscape
alteration operating in the Santa-Chao region during the last 5000 years.
Tectonic uplift, river mouth migration, and beach ridge formation all
necessitated adjustments by local human populations and are therefore
of particular archaeological importance.

Powered by tectonic forces, uplift stranded whole bays and altered
the physical landscape, the availability and location of coastal re-
sources, and hence the pattern of human settlement. We have found
evidence of two major episodes of uplift, the Salinas Event at about
5000 B.P. and the Chao Event at about 4000 B.P. The Salinas Event
coincides with the termination of a warm-water molluscan association
that indicates coastal climate conditions very different from the mod-
ern, cool-water regime. Because this uplift occurred just at the time
when sea level was stabilizing (Richardson, 1981), it preserved shell
middens and ancient beach deposits which might otherwise have been
submerged beneath the rising sea. This fortuitous preservation may
explain the lack of warm-water molluscs in other early strandline oc-
cupations and further suggests that the postulated shift from warm-
water to modern cool-water conditions was temporally if not causally
related to the stabilization of sea level.

Uplift also set the stage for subsequent beach ridge formation by
providing abundant sediment and a sufficiently straight coastline to

296

Annals of Carnegie Museum

VOL. 52

allow cobble deposition and ridge construction. By constantly pro-
grading the coast as well as by determining the physical characteristics
of the intertidal zone (thereby controlling intertidal resources), ridge
formation also affected human occupation. In the Pampa las Salinas
area, where beach ridges became prominent features following the uplift,
maritime-oriented occupation ended after the Salinas Event. In the
Salinas de Chao area, beach ridges are much less signihcant and marine
resources continued to be exploited long after both the Salinas and
Chao Events.

Beach ridge formation depends not only on the direct effects of uplift,
but also on the direction of the prevailing longshore current (to the
north), river mouth (sediment source) location, and quite probably on
the presence and intensity of rains accompanying major El Nino events.
If the El Nino correlation is correct, then the Santa beach ridges are of
tremendous potential importance in studying the history of the El Nino
phenomenon.

Finally, the Salinas Event probably triggered the southward migra-
tion of the Santa River mouth. A shift in the lower course of the river
meant a shift in the location of drinking water, irrigation water, and
the floral and faunal resources of the riverine microenvironment. Each
of these changes required adjustments by the local human populations.

Uplift, beach ridges, river mouth migration, changing climate, hu-
man occupations, alterations in settlement pattern, mollusc exploita-
tion, and changes in molluscan assemblages are all observable in the
Santa-Chao area. The correlations between these phenomena suggested
here still require rigorous testing in the held and in the laboratory, but
it is hoped that our observations will contribute to the growing aware-
ness of the importance of such processes in understanding coastal cul-
tures both in Peru and in other areas where similar processes occur.

Acknowledgments

Many of the ideas presented in this paper profited from numerous discussions with
other investigators of Peruvian archaeology and geology, especially Jack Donahue and
Michael E. Moseley. The idea of separate tectonic blocks separated by river systems
along coastal Peru was suggested to us by Michael E. Moseley. Thomas J. DeVries and
Thomas F. Lynch both made useful comments on earlier versions of the manuscript.
We are indebted to Robert L. Stuckenrath of the Radiation Biology Laboratory of the
Smithsonian Institution for running the radiocarbon dates. Field work in Peru was
supported by Fulbright and National Science Foundation Fellowships to Sandweiss and
by the Programa Riego Antiguo, the University of Pittsburgh Provosts Fund, and the
M. Graham Netting Research Fund of Carnegie Museum of Natural History. We grate-
fully acknowledge the assistance of all these individuals and organizations, though all
errors are, of course, our own.

1983

Sandweiss et al.-— Human Occupation on Peruvian Coast

297

Literature Cited

Alva Alva, W. 1978. Las Salinas de Chao: un complejo preceramico. Pp. 274-276,
in Actas y Trabajos del III Congreso Peruano del Hombre y la Cultura Andina, (R.
Matos M., ed.) Lima, 1:1-289.

Barazangi, M., and B. L. Isacks. 1976. Spatial distribution of earthquakes and sub-
duction of the Nazca plate beneath South America. Geology, 4:686-692.

Burnett, W. C. 1980. Apatite-glauconite associations off the coast of Peru. J. Geol.
Soc., 137:757-764.

Cardenas Martin, M. 1977-1978. Obtencion de una cronologia del uso de los recursos
marinos en el antiguo Peru. Arqueologia PUC, 19-20:3-26.

. 1979. A chronology of the use of marine resources in ancient Peru. Pontihca

Univ. Catoloica del Peru, Instituto Riva-Aguero, 104:1-30.

Craig, A. K., and N. P. Psuty. 1971. Paleoecology of shell mounds at Otuma, Peru.
Geographical Review, 61:125-132.

Daggett, C. 1983. Casma incised potter>': an analysis of collections from the Nepena
Valley. Pp. 75-97, in Investigations of the Andean past (D. Sandweiss, ed.), Cornell
Latin American Studies Program, Ithaca, 293 pp.

Grolier, M. J., G. E. Ericksen, J. F. McCauley, and E. C. Morris. 1979. The desert
landforms of Peru: a preliminary photographic atlas. U.S. Dept Interior, Geol. Surv.,
Interagency Report, Astrogeology 57:1-325.

King, C. 1973. Dynamics of beach accretion in south Lincolnshire, England. Pp. 73-
98, in Coastal geomorphology (D. R. Coates, ed.), Binghampton, New York,
326 pp.

Moseley, M. E. 1975. The maritime foundations of Andean civilization. Cummings,
Memlo Park, California, 1 3 1 pp.

Moseley, M., R. A. Feldman, and C. R. Ortlofe. 1981. Living with crises: human
perception of process and time. Pp. 231-267, in Biotic crises in ecological and
evolutionary time (M. Nitecki, ed.). Academic Press, New York, 343 pp.

Olson, A. A. 1961. Molluscs of the tropical Pacific. Paleont. Res. Inst., Ithaca, New
York, 456 pp.

Ortlofe, C. R., M. E. Moseley, and R. A. Feldman. 1982. Hydraulic engineering
aspects of the Chimu Chicama-Moche intervalley canal. Amer. Antig., 47:572-594.

PozoRSKi, S., AND T. PozoRSKi. 1979. Alto Salaverry: a Peruvian coastal preceramic
site. Ann. Carnegie Mus., 48:337-375.

. n.d. Almejas.

Richardson, J. B. III. 1981. Modeling the development of sedentary maritime econ-
omies on the coast of Peru: a preliminary statement. Ann. Carnegie Mus., 50:139-
150.

Rollins, H. B., J. B. Richardson III, and D. H. Sandweiss. n.d. New evidence and
implications of a major change in the structure of the east Pacific water mass about
5000 B.P.

Rollins, H. B., D. H. Sandweiss, and J. B. Richardson III. 1981. A thermally anom-
alous molluscan assemblage (TAMA) from archaeological sites along coastal Peru:
paleoecological implications and biostratigraphical utility. Abstracts, Geol. Soc. Amer.
Meeting, 13:540.

Sandweiss, D. H. n.d.a. Man and mollusc in prehistoric Peru. Unpublished Senior
thesis, Yale Univ., 1979, 124 pp.

. n.d. Origin of complex society on the Peruvian coast.

Sandweiss, D. H., H. B. Rollins, and T. Anderson. 1981. A single large magnitude
uplift in the Holocene record of the Peruvian north coast. Abstracts, Geol. Soc.
Amer. Meeting, 13:545.

298

Annals of Carnegie Museum

VOL. 52

ZiNSMEisTER, W. J. 1974. A new interpretation of thermally anomalous molluscan
assemblages of the California Pleistocene. J. Paleont., 48;84--94.

7. 73

ISSN 0097“4463

ANNALS

0/ CARNEGIE MUSEUM

CARNEGIE MUSEUM OF NATURAL HISTORY

4400 FORBES AVENUE « PITTSBURGH, PENNSYLVANIA 15213

VOLUME 52 16 SEPTEMBER 1983 ARTICLE 13

PRINCIPLES OF AGRARIAN COLLAPSE IN THE
CORDILLERA NEGRA, PERU

Michael E. Moseley^

Robert A. Feldman^

Charles R. Ortloff^

Alfredo Narvaez^

Abstract

The amount of land under cultivation on the arid north coast of Peru has decreased
by 35 to 40% during the last 1000 years. Two physical processes are implicated in this
agrarian collapse. The basic contributor}^ process is uplift along the Peruvian continental
margin, which causes ground slope changes, The second process involves rare but re-
current rains caused by El Nino perturbations of the normal marine and meteorological
conditions. Uplift leads to river entrenchment, and El Nino rains aggravate erosional
downcutting. Downcutting strands canal intakes, forcing abandonment of irrigation sys-
tems in favor of newer, lower canals which irrigate less land.

Introduction

The largest irrigation reclamation projects ever put into operation
in South America are sophisticated pre-Columbian canal delivery sys-

! ‘ Address: Curator, South American Archaeology, Field Museum of Natural History,

Roosevelt Rd. and Lakeshore Dr., Chicago, IL 60605.
i 2 Address: Visiting Assistant Curator, South American Archaeology, Field Museum of

j Natural History, Roosevelt Rd. and Lakeshore Dr., Chicago, IL 60605.

I ^Address: FMC Corporation, Central Engineering Laboratories, 1185 Coleman Ave.,

I Santa Clara, CA 95052.

Address: Universidad Nacional de Trujillo, Trujillo, Peru.

Submitted 22 March 1983.

299

300

Annals of Carnegie Museum

VOL. 52

terns designed to feed highland runoff from the Pacific watershed of
the Cordillera Negra to arid lands along the desert coast of Peru between
ca. 6° and 9° south latitude (Kosok, 1965; Schaedel, 1951). Built prin-
cipally after A.D. 500, but abandoned prior to European arrival, these
networks, discovered four decades ago with the advent of systematic
aerial photography, include multi-valley canal linkages (Kus, 1972;
Ortloff et aL, 1982; Eling, 1981; Shimada, 1981) as well as intra- valley
canal systems, remnants of which surround and enclose the 30 to 40%
smaller confines of contemporary irrigation agriculture (Moseley and
Deeds, 1982). Significantly, financing and technology imported from
industrialized nations are presently employed to erect large-scale rec-
lamation projects within the ruins of these abandoned networks without
knowledge of why the older and larger agrarian systems collapsed.

Within the lower Rio Moche valley (8° south latitude), between 1976
and 1980, 578 excavations were opened in abandoned canals and re-
lated agricultural structures (Pozorski and Pozorski, 1982). Hydraulic
analyses of channel configuration and engineering parameters indicate
that pre-Hispanic canal systems constructed during the past millen-
nium employed essentially modern hydrological concepts of critical
and sub-critical flow design techniques (Ortloff et al., 1982, 1983a,
1983Z?). This is not to imply either that native hydraulic engineering
is well explored, or that all abandoned canals are free of design errors.
However, these analyses support the contention that the roots of agrari-
an collapse do not lie with human error or technological inadequacies
of the societies which built the reclamation works (Ortloff et al., 1982,
1 983Z)). Proceeding from this position, it is possible to analyze irrigation
agriculture as an artificial extension of the natural hydrological regime
and to identify the long-term principles of agrarian collapse within the
Moche Valley and adjacent river basins.

Both regional processes have been implicated and local causes iden-
tified in the break-down of irrigation systems between 6° and 9° south
latitude (Ortloff et al., 1982; Eling, 1981; Shimada, 1981; Kus, 1972,
1983; Ericksen et al., 1970; Plafker, Ericksen, and Concha, 1971; Nials
et al., 1979; Moseley et al., 1981).

Steep-gradient rivers of the watershed are eroding in response to two
regional processes. The basic contributory process is uplift related to
high rates of tectonic activity along the continental margin, which can
cause ground slope change. The second process entails rare, but re-
current torrential rainfall on the otherwise completely arid coastal des-
ert caused by El Nino perturbations of the normal Pacific marine and
meteorological conditions. When the rare, major rains do fall— as last
transpired in 1925— it is upon steep, unvegetated land surfaces covered
with accumulated debris. As a result, the drainage system is flooded
and extensive and intensive erosion and mass wasting occurs.

1983

Moseley et al. — Peruvian Agrarian Collapse

301

Canal intakes physically tie irrigation to the drainage systems. As
rivers downcut and alter course, canal intakes loose efficiency and
become stranded above the entrenched stream channel. Intakes may
be reworked, but bedrock obstacles in the valley necks eventually curtail
recutting, leading to final intake stranding and channel abandonment.

Due to the lack of annual precipitation below ca. 1500 m, the lower
48% of the watershed lies within the world’s driest desert (Lettau and
Lettau, 1 978) so the water table of the coast land can only be recharged
by river runoff. As uplift and river downcutting proceed, the water table
falls and agriculture dependent upon groundwater constricts toward
the coast.

At the valley mouths, disgorged sedim.ent is sorted by constant north-
northwest longshore currents and distributed along the shoreline. Coastal
flats exposed by uplift provide abundant sand that is transported inland
by constant north to north-northwest daily winds. Sand drifts inundate
agricultural systems and force their abandonment. If they stabilize, the
sands may be farmed over and subsequently experience flooding, ero-
sion, or deflation, forcing renewed abandonment.

Locally, displacements along faults alter ground slope and change
canal gradients, thereby curtailing water supply to the distal reaches of
the channels. These tectonic events may lead to channel reconstruction,
new course placement, or canal abandonment (Ortloff et aL, 1983Z?).

This paper will focus on one of the above factors— river downcut-
ting—and the effects it had on irrigation and human settlement in the
Moche Valley.

Physical Factors
The Watershed

In concordance with adjacent drainages (Table 1), the Moche basin
rises a short distance inland at a high elevation (3988 m) and descends
the western slope of the Cordillera Negra at a very steep average gra-
dient (2.2 degrees) over a very short (102 km) linear course perpen-
dicular to the coast (Fig. 1). The steep river gradient and the often
precipitous bedrock landscape through which the river descends sug-
gests apparent ongoing uplift of the watershed, probably related to
tectonic processes at the active Peruvian continental margin, where
two lithospheric plates converge. The plate rotation model of Minster
et al. (1974) predicts that the Nazca oceanic plate is converging with
the South American continental plate at rates ranging from 8.9 cm per
year near the equator to 11.1 cm per year in central Chile. Deep sea
cores from the vicinity of the study area have been interpreted as
compatible with convergence rates of ca. 10 cm per year; such a rate
appears to have been the norm for the last 5 million years (Kulm et
al, 1976:795, 800).

302

Annals of Carnegie Museum

VOL. 52

Table \. — Characteristics of rivers of western Peru.

River

Basin
size (km^)

Irrigated
area (km^)

Length

(km)

Flow volume
(m^ X 10®)

150 m
elevation
(km from
mouth)

% Gradient
lOOO-Omasl

Chancay

3375

523.42

194

701

58

1.1

Zana

1125

191.13

119

202

51

1.0

Jequetepeque

4050

295.78

154

945

30

1.1

Chicama

3004

403.71

164

783

27

1.1

Moche

1562

200.26

105

320

19

2.1

Viru

1308

164.05

89

105

24

2.2

Chao

600

NA

78

NA

22

2.4

Santa

11,250

86.43

332

4594

27

0.9

The Moche basin and adjacent drainages between 6° and 9° lie within
the second of five longitudinal seismic provinces subdividing the An-
dean Cordillera (Barazangi and Isacks, 1976; Sillitoe, 1974). In recent
decades, seismic activity has included, in 1970, the most devastating
historic earthquake in the New World (Ericksen et ah, 1970; Plafker
et ah, 1971), as well as prior gradual vertical oscillation of the coastline
at an average rate of 1.8 cm per annum (Wyss, 1978).

Although the 1970 earthquake (which registered 7.7 on the Richter
Scale) was not known to be associated with surface fault displacement,
the watershed is cut by numerous geologic linears, many of which are
presumably geological structures. A pronounced scarp parallels the
Cordillera and subdivides the watershed into two longitudinal sections:

1. A coastal lowland consisting, not of marine sediments, but of
Quaternary and recent alluvium and aeolian sands overlying ig-
neous bedrock.

2. A broken highlands of highly dissected bedrock exposures with
limited cover of Quaternary and recent alluvium.

The pronounced scarp intersects the coast at a low angle in the region
between ca. 8° and 9° south latitude (Fig. 1), resulting in a wedge-
shaped coastal lowland. The major scarp is, in turn, cut by transverse
linears. Where these intersect the coast, inflections suggest offset and
the occurrence of related fault displacement.

Like the adjacent rivers, the Rio Moche’s valley coincides with a
transverse linear. In the highlands the river channel is entrenched with-
in a steep bedrock canyon. After crossing the scarp face, the flood plain
widens as the valley is not restricted by bedrock. Below ca. 150 m,
beyond the confines of the scarp, the gradient is lower and the river
traverses alluvial fill to the coast. As in adjacent drainages, the northern
side of the lower valley is characterized by mountain outliers which

1983

Moseley et al. — Peruvian Agrarian Collapse

303

CHANCAY

ZANA

CHICAMA

MOCHE

CHAO

SANTA

Fig. l.—Uncorrected mosaic of LANDSAT images showing the Peruvian north coast
from the Chancay drainage south to the mouth of the Santa River. Lines a-a and b-b
mark the high-flexure scarp that divides the coastal plain from the mountains.

crop out from the ocean from within large expanses of relatively flat
land, whereas south of the river the mountain fringe extends down to
the coast and there is relatively little flat land.

The lower 48% of the Moche watershed lies in a hyperarid desert
(Lettau and Lettau, 1978), where normally there is no annual precip-
itation below 1 500 m. Infrequent desert rains occur only in association

304

Annals of Carnegie Museum

VOL. 52

with strong El Nino perturbations of the normal ocean-atmosphere
conditions, which have a statistical periodicity of about one per 1 5 or
16 years in the region of the Rio Moche (Nials et al., 1979). However,
not all strong perturbations produce showers, so major rains— such as
last occurred in 1925 — may fall less than once per century or two.
When a major deluge occurs, it is upon a steep, un vegetated landscape
covered with accumulated debris. If, during the decades or centuries
between major rains, the landscape experiences even minor ground
slope alteration from seismic or gradual tectonic activity, then all of
the normally dry desert drainages will be out of erosional equilibrium.
When the destabilized hydrologic regime is supercharged by flooding,
erosion and mass wasting can occur on a scale so vast that Holocene
incidents, as last occurred ca. A.D. 1 100, have been misidentified as
products of earlier glacial epochs (Moseley et al., 1981).

Only the upper 52% of the Moche basin collects significant precip-
itation. Highland rainfall is markedly seasonal, with more than 75%
falling in 25% of the calendar year. River flow peaks between February
and March, then drops rapidly. For three-quarters of each year most
river discharge is drawn off by irrigation agriculture in the lower valley,
where there are ca. 19,950 ha of land under cultivation, most of which
lies below 1 50 m.

The water table of the coastal lowlands, charged by river runoff, has
a high-low cycle peaking between June and September, well after the
river has crested. Groundwater provides ca. 1 6% of the present coastal
agricultural water supply (ONERN, 1973). It is exploited via chan-
nelized springs or pukios, pumps, and sunken gardens. The latter are
an indigenous technique where the land surface is excavated down to
the level at which natural soil moisture can sustain plant growth. Sunk-
en gardens are an efficient agricultural technique, although one restrict-
ed to areas with a high water table by the large labor expenditures
needed for excavation.

The Moche basin and its agricultural area fit within a graded north-
to-south sequence of large to small river valleys with agricultural areas
of proportional size (Table 1; Fig. 2). This pattern begins with the Rio
Chancay (6.5° south latitude) and, with one exception (Rio Zafia) con-
tinues south to the Rio Santa, where it is broken. The Santa is the
largest river basin along the entire Andean desert coast, and while it
carries 14 times the volume of the Rio Moche, it supports a significantly
smaller agricultural area (43%).

The Hydrological Regime

If irrigation agriculture is to be analyzed as an artificial extension of
the hydrological regime, then it must be recognized that this regime
evolves in a mechanical manner in response to changes in land-to-sea
level at the river mouth and along the littoral zone. Due to eustatic

1983

Moseley et al.™ Peruvian Agrarian Collapse

305

306

Annals of Carnegie Museum

VOL. 52

and tectonic changes, the littoral zone, the river mouths, and the entire
hydrological regime of the Cordillera Negra have never stabilized dur-
ing the course of human occupation. With the onset of glacial meltback
some 15,000 years ago, the level of the oceans rose an estimated 85 to
135 m and, in the course of 10,000 years, submerged the continental
shelf, including more than 75 km of once-exposed coastal plain at the
mouth of the Moche Valley (Richardson, 1981; Moseley, n.d.). Rising
sea-to-land levels put river mouths into aggradational regimes and
generated high groundwater conditions inland of the submergent coast-
line.

Although the ocean level began to stabilize approximately 5000 years
ago, the watershed did not. Rather, there was a reversal of hydrological
conditions because uplift of the watershed remained an ongoing proces.
Whereas the sea rose faster than the land up to ca. 3000 B.C., the land
has risen relative to the sea since then. Rising land-to-sea levels put
rivers into a degrading regime and resulted in lowering of the ground-
water table inland of the emergent coastline.

The change from a rising sea level to a rising land level is demarcated
by a prominent wave-cut sea cliff that is largely continuous for the
length of the coast between ca. 6.5° and 9°. The association of this cliff
with the Holocene shift from submergent to emergent regimes is in-
dicated by the absence, inland from it, of any Tertiary or Quaternary
marine deposits or coastline features (Cossio and Jaen, 1967). Further,
the earliest sedentary human communities to occupy the top of the
cliff are radiocarbon dated to between 3200 and 2000 B.C. (Carde-
nas, 1977-78; Bird, 1951; Pozorski and Pozorski, 1979). The earliest
mollusks, occurring both as stranded colonies on uplifted beach sur-
faces adjacent to the cliff base and in the cliff-top midden, date to ca.
3200 B.C. (Sandweiss et al., 1981). No human occupation prior to this
date has been found below the sea cliff. The earliest human occupation
tentatively associated with the highest (8 m) marine terraces at the
mouth of the Moche Valley is dated, by ceramic style, to the latter half
of the first millennium B.C. (Nials et al., 1979).

The contemporary littoral zone is seaward of the cliff, with inter-
vening distances grading systematically from a maximum of 5 km near
the mouth of the Rio Santa (9° south latitude) to a minimum of a few
tens of meters at Puerto Eten, near the Rio Chancay (6° south latitude),
indicating a north-to-south increase in shoreline width (Sandweiss et
al., 1981). There also appears to be a north-to-south increase in uplift;
no terracing is present near the mouth of the Rio Chancay (Shimada,
1981), whereas the highest terrace seaward of the cliff at the Rio Moche
is 8 m. high (Cossio and Jaen, 1967), and uplifted terraces and beach
ridges have their maximum reported height of 1 5 m near the Rio Santa
(Sandweiss et al., 1981).

1983

Moseley et al. — Peruvian Agrarian Collapse

307

Although the sea cliff is largely continuous across the mouth of the
Moche Valley, marine terracing and beach ridges are found in their
most developed form on the northern, leeside of bays bracketing the
valley mouth. The bays only have seaward projecting headlands on
their south sides and they represent fault-bounded coastal deflections.
The distribution of well developed raised beach surfaces in these set-
tings may reflect consistent relative uplift of the south side of the faults.

The River Banks

In recent millennia the Rio Moche, below 150 m, has both downcut
and cut laterally, principally southward. This movement has resulted
in an asymmetrical erosional profile, with gentle slopes north and west
of the river and a vertical bank formed by undercutting of valley-floor
deposits on the south and east side of the river. Beginning ca. 6.5 km
in from the river mouth and continuing upstream for more than 7 km
to the valley neck, undercutting has exposed a largely continuous strati-
graphic profile 10 m or more in height. Salient features of a 12+ m
column (Fig. 3) near the profile’s seaward end include:

A) Basal peat deposits that are radiocarbon dated to ca. 1000 B.C.
(WSU No. 2190: 3010 ± 100 B.P.; WSU No. 2194: 3020 ± 60
B.P.) (Fig. 3h, i).

B) Ceramic inclusions in fluvial deposits overlying the peats that
are dated on stylistic grounds to no later than ca. 500 B.C. (Fig.
3f).

C) Buried agricultural furrows and a small sherd-lined feeder canal
at the top of the column, dating to the Early Chimu phase prior
to ca. A.D. 1000.

The top of the column has been deflated, and behind the bank are
yardangs, or butte-like erosional remnants of sandy loam, which stand
several meters high and contain Early Chimu sherds. Early and Middle
Chimu occupational debris are found on the deflated surfaces between
the yardangs. Thus, the end of deposition and the onset of erosion can
be dated to within the Early Chimu phase (Table 2).

The onset of erosion is the postulated product of river downcutting
in response to tectonic activity. Unfortunately, the stratigraphy of the
south bank sequence does not crop out downstream as far as the littoral
and therefore cannot be tied to a specific uplifted beach surface.

The Water Table

Because river flow charges groundwater, uplift with consequently
increased gradients and river downcutting would, expectably, alter water
table conditions. Such a change in groundwater conditions can be seen

308

Annals of Carnegie Museum

VOL. 52

SAND

SILTY-SAND

MARL

SILT

PEAT

Fig. 3. — Rio Moche stratigraphic column at Pampa Casique. a) Chimu fields; b) Chimu
feeder canal; c) buried fields; d) hearth; e) 2180 ± 170 B.P. (WSU-2192); f) pottery
sherds; g) 2490 ± 90 B.P. (WSU-2193); h) 3020 ± 60 B.P. (WSU-2194); i) 3010 ± 100
B.P. (WSU-2190).

1983

Moseley et al. — Peruvian Agrarian Collapse

309

Table 2.

PERUVIAN

RELATIVE

CHRONOLOGY

LOCAL

PHASES

MAJOR SITES

1500-

1000-

500—

OAD—

LATE HORIZON

CHIMU-INCA

LATE

INTERMEDIATE

PERIOD

LATE BRICK

MIDDLE BRICK

CHIMU

EARLY BRICK

CHAN CHAN

MIDDLE

HORIZON

^ EARLY CHIMU

V

1

GALINDO

EARLY

INTERMEDIATE

PERIOD

IV

III

MOCHE

II

1

MOCHE HUACAS

in the horizontal distribution of two agricultural techniques— spring
fed canals and sunken gardens.

The earliest historical map of the valley’s agricultural configuration,
made in 1760, shows two spring fed canal systems north and west of
the river (Kosok, 1965). The highest system was fed from a pond located
about 75 m above sea level, 12 km inland, and 1.5 km north of the
river. By 1942, when aerial photographs of the lower valley were made,
this system was no longer operative, and the area it once irrigated was
supplied by the major northside river fed canal (the Nl). The lower
canal system fed by groundwater was shown in the 1760 map as coming
from a pond located about 42 m above sea level, 9 km inland, and 2
km north of the river. By 1942, this system (the Pukio system. Fig. 4)

310

Annals of Carnegie Museum

VOL. 52

was no longer fed from a pond, but rather from two springs or seeps
approximately 1 km apart. Today it is fed only by the lower spring.

Sunken gardens, which rely upon high groundwater conditions, ex-
tended inland at least 4.5 km at the beginning of the Chimu phase (Fig.
4). By 1942, this technique had dramatically contracted coastward and
all sunken garden farming was limited to a region within 1 km or less
of the shoreline, principally seaward of the wavecut clilf. Areas formerly
cultivated by groundwater exploitation were irrigated from river and
spring fed canals. Coastward encroachment of irrigation agriculture
into these areas has continued in recent decades without the significant
salinization problems that would be expected if the water table had
not subsided, improving drainage. A measure of this subsidence is
reflected by remnants of abandoned early Chimu gardens situated 4.5
km inland that now lie 10 to 12 m above the contemporary water table.
Although this drop is a cumulative measure that includes the effects
of mechanical pumping, the seaward contraction of groundwater ag-
riculture which preceded pumping is most simply explained as a cor-
ollary of coastal uplift, river downcutting, and water table lowering.

Canal Strategy

In a context of high aridity for unlined canals, less than 55% of the
water that a canal receives from its river source reaches farmland, less
than 45% is absorbed by the soil, and only about 1 6% is actually utilized
by crops (USDA, 1955; West, 198 1). Much of the loss occurs as seepage
or evaporation during transport. Therefore, the highest system effi-
ciency in open channel flow design entails moving the maximum amount
of water the minimum possible distance to the largest possible planting
area.

For the steep-gradient rivers of the Cordillera Negra, maximum de-
sign efficiency is represented by canals that transport and distribute
water along a course perpendicular to the rivers. This configuration
expresses a balanced relationship between separate functions of supply
and distribution. Intake capability, slope, and channel hydraulic con-
figuration set maximum limits on the volume of water that a canal can
supply. However, the maximum limit on the size of the planting area
over which this supply can be efficiently distributed is the downslope
area included between the river and the canal. This “angle of reach”
between river and canal thus establishes the distance of transport rel-
ative to area of planting surface (Fig. 5).

If a canal intake feeds from an entrenched river, then its initial lead-
off channel must slope downstream at a low angle of reach before it
can exit the confines of the downcut floodplain at an elevation suitable
for perpendicular course placement. Given water loss in transport, the
perpendicular course distance that the canal can supply is thus limited

MOCHE VALLEY

311

1983

Moseley et al. — Peruvian Agrarian Collapse

Fig. 4. — Map of the lower Moche Valley showing major canals, pampas, and areas of abandoned fields.

312

Annals of Carnegie Museum

VOL. 52

RIVER

I

SLOPE, ANGLE, AND OPTIMUM STRATEGY

Fig. 5. — Irrigation parameters. The optimum strategy is to irrigate the maximum amount
of land for a given canal length, thus minimizing water loss during transport to the fields.

by the length of the low-angle lead-olf channel, and the lead-off distance
is relative to the depth of river entrenchment.

Supply to the perpendicular course will decrease if the canal mouth
and lead-off channel must be reworked in response to ongoing river
downcutting, which strands the intake above the entrenching stream
flow. By reworking and extending the stranded end of the lead-off
channel upstream from its original position and having the prolonged

1983

Moseley et al.-=- Peruvian Agrarian Collapse

313

lead-ofF channel intersect the river at a higher elevation, a new intake
can be constructed which maintains the canal supply, but at the ex-
pense of increasing low-angle transport distance. In other words, as the
lead-olf channel is extended farther and farther upstream, it more
nearly parallels the river, leading to a locally small angle of reach
between the canal and river. As downcutting continues, upstream ex-
tension can be repeated, and the intake will “migrate” upriver until
eventually a bedrock obstacle is encountered.

In the narrow bedrock valley neck, the task of trenching or tunneling
through solid rock curtails lead-ofF extension, while aqueducting around
the valley neck obstacles is inhibited by lateral river movement, which
undercuts the canal. When bedrock abutment of the intake occurs,
supply can be maintained only by lowering the intake to the river level
by trenching the intake and lead-off channel to a lower slope. However,
decreasing channel bed slope for a fixed canal cross-sectional area de-
creases the total supply flow rate.

When hydraulic efficiency of the lead-ofF channel is lost through
upstream migration of the intake, or its lowering at bedrock abutments,
there remains an option of constructing one or more new downstream
intakes that connect into the original channel along its lower course,
thereby augmenting the total supply. However, the point at which the
canal swings out perpendicular to the river sets a limit on additional
intakes. Beyond this point, recharge cannot occur, nor can downriver
canals reach fields along the perpendicular course of the atrophying
higher system.

In combination, bedrock abutment and elevational strictures on re-
charge from multiple intakes set finite limits on adjustment to ongoing
river entrenchment, and thus on sustaining irrigation from perpendic-
ular-reach canals at any given elevation. When these limits are reached,
the supply capacity of the canal diminishes below the needs of the
planting area encompassed within the angular reach of the canal. Due
to water loss in transport, sustaining the farthest fields is uneconomical,
so cultivation retreats upchannel, constricting back towards the river
at a rate relative to the decreasing intake capacity.

Bedrock abutments at the valley neck inhibit building a new “re-
placement” canal system upstream from the old one, and thus promote
new construction downstream, where there is sedimentary fill and room
for intake adjustment in response to further river changes. In a context
of continuing entrenchment, this fosters successive downstream con-
struction and replacement of canal systems. Through time, irrigation
agriculture retreats coastward at descending elevations in a step-like
manner. Entrenchment eventually reaches the stage where the distance
between the intake and the coast is insufficient for lead-off channels to
pull out of the downcut landscape and swing laterally to a perpendicular

314

Annals of Carnegie Museum

VOL. 52

course placement. Thus, in overview, the angle of reach and area of
irrigation agriculture in any given valley are relative to its degree or
depth of river entrenchment. The Rio Santa, which has an extraordi-
narily large discharge volume, supports an extraordinarily small agri-
cultural area because it is the most deeply entrenched river on the north
coast.

Historical Perspective

The physical record of how irrigation systems behave through time
is not unlike the record of glacial behavior. As ice or agriculture advance
over a landscape, evidence of earlier advances or contractions is largely
destroyed. However, as the ice melts or agriculture contracts, they leave
behind moraines or abandoned fields and canals as evidence of their
retreat. While there may have been more than one early agricultural
advance in the Moche drainage, most abandoned agrarian structures
of easily recognizable form date later than ca. A.D. 500 and pertain to
two pre-Hispanic occupations— Moche and Chimu (Table 2). During
the later occupation, the large metropolitan center of Chan Chan, north
of the river, became the imperial capital of Chimor, a desert polity
with political hegemony reaching over the coast from southern Ecuador
to central Peru. Agricultural lands surrounding the city were worked
and administered as a unified plantation system (Moseley and Deeds,
1982; Keatinge and Day, 1973; Keatinge, 1974). Pre-Hispanic agri-
cultural works survive, in analyzable form, outside the area of contem-
porary farming, and therefore reflect agrarian collapse at the distal ends
of the irrigation system. At the entrenching river, evidence of aban-
doned intakes and lead-off adjustment has been destroyed by river
movement or masked by continuing cultivation.

The Canal Configuration

The river-fed irrigation system of the lower Moche drainage is based
on three Primary Maximum Elevation Canals (PMECs) on each side
of the river (Fig. 4). Maximum Elevation Canals are channels having
the highest elevation in a particular section of the valley, and thus
delimit the angular reach and irrigated area. These canals are numbered
sequentially upstream and are lettered N or S, indicating whether they
are north or south of the river. Although potential for angular reach
increases in the upstream direction, the area currently served by each
successively higher PMEC decreases (Table 3).

Local farmers indicate that within the span of living memory canal
intakes have been a continuing source of problems. The SI intake,
which is bedrock abutted, has both an abandoned cement mouth and
a higher, more recent, intake and flow diversion structure of rustic
wood and cobble construction. The N2 intake has migrated upstream

1983

Moseley et al. — Peruvian Agrarian Collapse

315

Table ?>. — Characteristics of modern and ancient Moche Valley canals.

Canal

Capacity .
(m^)

Modern area served

Ancient maximum area

Ha

%

Ha

%

N3 Moro

2.5

700

6.7

1509

8.2

N2 Vichansao

3.0

1363

13.1

5821

31.6

N 1 Mochica

10.0

4859

46.6

8992

48.8

Pukios

1.4

1504

14.4

~

-

S3 Huatape

0.6

258

2.5

309

1.7

S2 Santo Domingo

1.5

759

7.3

1793

9.7

S 1 General de Moche

2.0

987

9.5

—

—

Totals

21.0

10,430

100.1

18,424

100.0

since the turn of the century and now has merged with the N3 intake
at a valley neck bedrock abutment. Although both canals feed off the
same intake and have only slightly different maximum capacities (2.5
versus 3.0 mVsec), the N3 is subject to silt clogging and irrigates only
about half the area of its counterpart (ONERN, 1973). This condition
would be expectable if the abutted N3 has experienced slope lowering
and thus lost transport efficiency as a result of intake lowering prior to
the N2’s recent migration to the abutment.

Channel sections of operational PMECs intrude upon archaeological
ruins of known antiquity, and therefore must date later than the struc-
tures they cut or the deposits they overlie. The S3 overrides Moche
phase III deposits, whereas the N3 overlies phase IV materials but is,
in part, functionally associated with the site of Galindo, a phase V
settlement near the valley neck (Bawden, 1982). The operational N2
transects a ruin dating to the first millennium B.C., and an abandoned
section of it cuts through a Moche phase IV mound (Pozorski, 1982).
The S2 cuts Moche phase III-IV roads (Beck, 1979). Operational and
abandoned sections of the N 1 intrude upon Middle Chimu phase ar-
chitecture at Chan Chan (Kolata, 1982; Lange, 1971), whereas the SI
lead-off channel crosses a surface created by 17th century mining of
Huaca del Sol, a huge Moche pyramid, and therefore dates to the
Colonial period or later (Topic, 1982; Hastings and Moseley, 1975).

All of the operational PMECs, except the S 1 , have abandoned courses
extending out into what today is desert (Fig. 4). Abandoned extensions
and field areas of the N3 and S3 are less well preserved than comparable
aspects of the N2 and S2 systems, while abandoned N 1 extensions and
fields are in the best state of preservation.

Canal courses, such as those of the N3 and N2, converge and cross,
and excavations demonstrate that abandoned courses are often occu-
pied by multiple, superimposed channels. This situation may reflect

316

Annals of Carnegie Museum

VOL. 52

both the resupplying of older courses and the intentional use of the
sediments accumulated in beds of older channels as low-permeability
linings to inhibit seepage. When such courses cross or merge, the series
of superimposed channels may split and recombine in ways that defy
surface detection. Much of the excavated information comes from
channel transects downstream of unexplored course convergences and
therefore does not securely monitor flow routing along potentially al-
ternative paths. Thus, excavated transects must be approached as a
series of isolated profiles that: a) establish maximum potential water
supply vis-a-vis hydraulic limitations of channel design; and b) monitor
realization of supply potentials vis-a-vis bedload sediments and soil
oxidation, the presence of which establishes use, but the absence of
which does not disprove use.

Canal Classes

Survey, excavation, and analysis of hydraulic design and engineering
features allow recognition of three major classes of canal remnants
(Ortlolf et al., 1983Z?) (Tables 4-6).

Class 1 canals have unlined parabolic channel cross-sections typical
of the equilibrium shapes obtained after flow-induced sidewall erosion
has occurred in channels dug into loosely consolidated soils. These
channels supported subcritical flows and reflect hydraulic design mech-
anisms cognizant of that flow regime. Canal beds are oxidized from
long or intensive use, generally exhibit extensive silt accumulations,
and may underlie later canals built in or along the same course. This
class includes last-use extensions of the N3 and the S3, and remnants
of unconnected oxidized channels that originated higher up than the
present N 1 or S 1 and that pertain either to early phases of the higher
extant PMECs or to different, more ancient, systems.

Class 2 canals are generally trapezoidal in cross-section and are ma-
sonry lined with cobbles set in adobe mortar. These channels supported
either subcritical or critical flows by design manipulations of bed slope,
channel cross-section, and sidewall roughness. Oxidation is rare to
absent, and while unused courses are present, silt accumulations in-
dicate limited use occurred. This class includes last-use extensions of
the N2, as well as the S2, which saw several sequential phases of
masonry constructions that often overlie Class 1 channels.

Class 3 canals have unlined earth bank channels with profiles cut to
trapezoidal form but eroded to a parabolic shape. Flow is subcritical,
channel oxidation may occur, and bed sediment layers indicating use
are present. This class includes the last and present phases of the N 1
and the operational SI, as well as parts of the N2, N3, S2, and S3
canals.

Table — Characteristics of the three major classes of canal remnants in the Moche

Valley, Peru.

317

1983 Moseley et al. — Peruvian Agrarian Collapse

^ Underlined/slashes denote terminal phases.

318

Annals of Carnegie Museum

VOL. 52

Table 5. — Characteristics of canals on Pampa Huanchaco, Moche Valley, Peru.

Phase

Canal

class

Supply

canal

Maximum

flow

rate

(mVsec)

Comments/use indications

1

1

N3

and/or

1.8

All branches used.

2

1

N2

4.8

All branches used.

3

2

N2

2.7

Not all constructed branches used.

4

2

N2

1.3

Length of used canals contracting.

5

2

N2

1.1

Far upstream segments used, abandonment
of downstream segments.

Present

-

-

0

Total abandonment of Huanchaco system.

The pre-Hispanic water supply to the valley’s northwestern plains
can be studied in the N2 (Vichansao) profile from the excavation far-
thest upchannel and closest to the modern N2 (Table 6). Of four su-
perimposed PMEC channels, the earliest is a Class 1 canal which had
an estimated flow rate of 9.6 mVsec and which shows evidence of
substantial use. The first superimposed channel had roughly half this
capacity, a change that may reflect short-term alternative routing, as
this second channel was replaced by a third, larger, Class 2 masonry
channel that had an estimated capacity of 1 0.2 m^/sec. After destruction
by flash flooding around A.D. 1100 prompted reconstruction of the
canal, the final channel was built. Its flow rate was 3.1 mVsec, which
is essentially the same as the contemporary N2. The greatest present
flow capacity is that of the contemporary Nl, which can carry 10.0
m^/sec. Therefore, if the N 1 replaced the N2 after flooding and recon-
struction, then the long-term PMEC sequence does not reflect, in design

Table 6. — Characteristics of canals on north side of Rio Moche, Peru.

Canal/phase

Canal

class

Maximum

flow

rate

(mVsec)

Comments/use indications

G Canal

1

1.1

Fed by N3; use indicated.

Moro/ 1

1

2.6

N3; all branches used.

Moro/2 .

1

5.8

N3; all branches used.

Vichansao/ 1

1

9.6

N2; Early Class 1 type.

Vichansao/2

2

5.0

N2; A-Complex; all later phases show use.

Vichansao/3

2

10.2(?)

N2; A-Complex; all later phases show use.

Vichansao/4

2

3.1

N2; A-Complex; all later phases show use.

Intervalley

2

4.7

Flow rate theoretical: no indications of use.

Mochica

3

10.0

Nl; in use at present.

1983

Moseley et al.-— Peruvian Agrarian Collapse

319

intent or flow capacity, a lack of water or decline in its availability.
However, canal profiles at the distal end of the irrigation system do
reflect decreased delivery of water. On Pampa Huanchaco, at the west-
ern terminus of irrigated land, the first three channels of a five-phase
sequence reflect decreasing flow, whereas in two post-flood channels
there was a marked decrease in flow rate (Table 5) (Ortloff et al., 1 983Z?).

The Distal Canal Sequence

The largest area and longest reach achieved by well-preserved PMECs
north of the river are on the western plains of Pampas Esperanza, Rio
Seco, and Huanchaco, on the first of which Chan Chan was located.
Past and present N 1 courses cut into middle phase Chimu architecture
at Chan Chan, including the city’s north wall (which crosses and blocks
feeders from the uphill Class 2 channels) and a walled road that was
functionally compatible with a Class 2 extension of the N2 at the head
of Pampa Esperanza. This Class 2 channel overlies a Class 1 canal, and
both the wall and walled road override other Class 1 canal remnants
(Beck, 1979). From these and other stratigraphic superpositions, it is
evident that by Moche phase V times Pampa Esperanza was irrigated
via a Class 1 extension of either the N3 or N2 and that the system had
achieved an approximate perpendicular reach.

The Chimu inherited and reworked this system using Class 1 exten-
sions of the Esperanza MEC in order to reclaim the two western-most
pampas (Fig. 6a), which radiocarbon dates indicate was under irrigation
by about A.D. 1050. The entire Class 1 canal system, which shows
substantial use, was subsequently remodeled and the earlier earth-bank
canals were replaced with Class 2 masonry-lined channels (Figs. 6b,
6c). In some cases, these new channels had a trapezoidal cross-section
that minimized flow resistance and thus maximized the flow rate for
the cross-sectional area. This Class 2 system, which can be traced
intermittently back through the Vichansao’s profile to the operational
N2, saw little use and may still have been under construction at its
distal end on Pampa Huanchaco when massive El Nino flooding— of
a magnitude as yet unrepeated — occurred. This event, which transpired
around A.D. 1 100, eroded away more than a meter of valley floor along
the river, washed out irrigation works in all parts of the valley, and
adversely affected most of the north coast (Nials et al., 1979; Shimada,
1981).

The destroyed Class 2 system was reconstructed, but with signifi-
cantly lower flow rates. Throughout the western plains the reconstruct-
ed system saw little, then ultimately no, use. When Chan Chan’s 8 km-
long north wall was built across Pampas Esperanza and Rio Seco,
formerly farmed lands north and west of the city were left without

320 Annals of Carnegie Museum vol. 52

GROWTH OF CHAN CHAN AND ADJACENT AGRICULTURE

Fig. 6. — Growth of Chan Chan in relation to irrigation on Pampas Esperanza, Rio Seco,
and Huanchaco. Dots represent field areas, marsh symbol represents areas of sunken
gardens, and the rectangles represent the major compounds at Chan Chan (infilling shows
the ones in use at each period). A) First compound was built between irrigated fields and

1983

Moseley et al.~ Peruvian Agrarian Collapse

321

water. However, a major intervalley Class 2 canal, more than 70 km
long, was cut to deliver Rio Chicama water to the head of Pampa
Esperanza, and thereby resupply the dry N2 extension (Fig. 6d) (Kus,
1972; Day, 1974; Ortlolf et ah, 1982). A walled thoroughfare was
ceremoniously built from Chan Chan out to the junction of the Inter-
valley and N2 canals and downstream Class 2 feeders were refurbished
in order to distribute the intervalley water. However, masonry lining
of the main intervalley course was completed only as far as an active
fault north of the intervalley divide. Displacements along this fault,
which coincide with a prominent geologic linear, are indicated by
disrupted stream drainage patterns. Where the masonry-lined channel
approaches the fault, it rests directly atop a bedrock block that has
been upthrown at the fault. Today, this final segment of the finished
Class 2 Intervalley Canal channel has experienced a 1 degree uphill
slope change from its slope at the time of construction (Ortloff et ah,
1982).

Radiocarbon dates indicate that significant agricultural activity on
the western plains had ended by about A.D. 1350. The inland expansion
of Chan Chan ended with the abortive intervalley project, and Late
Chimu city construction moved sequentially coastward, back into the
old urban core near the beach (Fig. 6e). This retreat is understandable,
as the city depended on large, open wells for its potable water. When
irrigation ceased to recharge the aquifer that these wells tapped and to
mitigate seasonal water table fluctuations, urban growth was forced to
reverse itself and follow the declining groundwater levels coastward to
more accessible locations, in much the same manner as sunken garden
farming retreated coastward (Moseley and Kolata, 1982; Moseley and
Feldman, 1982).

The final step in the evolution of irrigation on Pampa Esperanza is
represented by Chimu construction of the N 1 PMEC and by westward
extensions of Class 3 canals across the lower half of the plain. At one
point, the N1 watered fields immediately northeast of the city wall,
and an unsuccessful attempt was made to channel water along the
upslope side of this wall. Next, a viable canal was cut through the wall

sunken gardens. B) Expansion of irrigation upslope on Pampa Huanchaco. C) Unification
of the irrigation system at its maximum extent. D) Expansion inland of Chan Chan,
partial abandonment of the intravalley system, and construction of the Chicama-Moche
Intervalley Canal. E) Abandonment of Pampa Huanchaco after failure of the Intervalley
project, Moche Valley N 1 canal built, and irrigation inside some of the older compounds
at Chan Chan. F) Modern system, with areas of former sunken gardens now farmed by
pump irrigation.

322

Annals of Carnegie Museum

VOL. 52

along a course that is today paralleled — at a lower elevation — by the
contemporary Nl. Fields and feeders were cut into middle and late
phase architecture in the northern part of Chan Chan, but not into its
densely occupied coastal sector. Today, some agriculture still goes on
among the city’s ruins (Fig. 6f)-

Parallel events south of the river are understood in less detail. Early
remnants of Class 1 systems are crossed by the last-use Class 2 channel
of the S2, which apparently was a composite post-flood reconstruction
of an earlier operational channel and which saw little use in its final
form. There is a hiatus between abandonment of the S2 (by or during
the Middle Chimu phase) and the Colonial period construction of the
SI. However, this late canal required landscape alterations that would
have obfuscated any prior SI variants that might have been present
during the Late Chimu phase.

Limitations on the Data

Data from the Moche Valley support the hypothesis that agrarian
collapse is an ongoing process principally related to uplift and ground
slope change generated by high rates of tectonic activity along the
continental margin, but punctuated by rare but recurrent El Nino flood-
ing. It must be understood that the hypothesis of agrarian collapse has
both geological components pertaining to the nature and rates of ground
slope change, and hydrological components related to the mechanical
consequences of river downcutting. At this stage of investigation, we
stress that the latter are better documented and more fully understood
than the former.

There is unequivocal evidence of river downcutting and course change
in recent millennia. These changes correlate chronologically with both
diminution of canal water supplies and contraction of ground water
agriculture. Indeed, the latter are expectable hydrological and agro-
engineering consequences of the former. River entrenchment can be
triggered by a variety of independent processes ranging from defores-
tation, through minor shifts in precipitation patterns, to coastal uplift.
Quaternary terraces and a downcut river profile can be traced through
the neck of the Moche Valley downstream to within ca. 6.5 km of the
river mouth. In this area the river crosses a pronounced linear (Fig. 1:
a-a) associated with the major scarp running between ca. 9° and 6°
south latitude. Below this crossing, the river profile has not been suc-
cessfully traced to the coast. Therefore, river downcutting has yet to
be stratigraphically tied to an uplifted beach surface, and alternative
triggering mechanisms for entrenchment cannot be ruled out.

It is generally conceded that the present-day north coast marine and
meteorological regime, which has communication with the broader

1983

Moseley et al.~ Peruvian Agrarian Collapse

323

tropical Pacific ocean-atmosphere circulation system, has been in place
for the last five millennia (Richardson, 1981; Rollins et a!., n.d.). El
Nino perturbations are an integral feature of this climatic regime, and
their occurrence in the archaeological record of recent millennia is
well documented (Nials et ah, 1979; Shimada, 1981). Thus, we think
it questionable that the greater than 30% decrease in irrigated land
during the last 1500 years is a product of climatic change or long term
alteration of precipitation patterns triggering river entrenchment.

Agrarian abandonment in the rainfall zone of the highlands has yet
to be quantified or investigated in detail. Our examination of relevant
aerial photographic coverage suggests a disparity between past and
present acreage of roughly comparable magnitude to that of the coast.
Within the zone of annual precipitation, much — or perhaps most —of
this abandoned land was farmed with the use of small canal systems.
In theory, if such land supported a more erosion-resistant vegetation
cover when under cultivation than when left abandoned, then a cor-
ollary of highland irrigation collapse could be “deforestation” and con-
sequent erosion that could exacerbate river downcutting, leading to
intake stranding of the coastal canal systems.

We believe that tectonic activity and ground slope alteration provide
a much simpler and more unified explanation of coastal and highland
agrarian collapse, as well as erosion and river downcutting. As discussed
above, the northern watershed of the Cordillera Negra is crossed by a
high flexure scarp (Fig. 1). Where this structure emerges from the sea,
at 9° south latitude, cumulative coastline displacement securely dated
to after ca. 3000 B.C. measures 5 km horizontally and more than 10
m vertically (Sandweiss et ah, 1981). In the same area, recent gradual
vertical oscillation has averaged 1.8 cm per annum (Wyss, 1978). We
suggest that these data constitute a record of structurally unified move-
ment transpiring over the length of the scarp northward to 6° south
latitude. From south to north the scarp trends inland and progressively
farther back from the sea. There is a corresponding decrease in hori-
zontal separation between the modern littoral and the wave-cut sea
cliff (which formed ca. 3000 B.C.), as well as a general vertical decrease
in the height of uplifted sea floors and terraces (Cossio and Jaen, i 967).
This could reflect a south to north decline in tectonic displacement.
Alternatively, as the scarp pulls inland the coastal record of activity
along its axis may simply become more indirect. In either case, there
is evidence of past— as well as ongoing— tectonic activity in the study
area. The same cannot presently be said of potential alternative causes
of river downcutting.

In considering these limitations on the data, we should note the
growing archaeological documentation of abandoned canal sections
that presently run uphill or have gradients that would be inoperable

324

Annals of Carnegie Museum

VOL. 52

today (Ortloff et al., 1982, 1983(2; Kus, 1983; Pozorski and Pozorski,
n.d.), as well as monumental architecture exhibiting progressive, cu-
mulative slope and orientation alterations transpiring over long periods
of occupation (Donnan and Ortlolf, 1983). The obvious advantage of
the tectonic hypothesis is that it readily accounts for these local phe-
nomena, as well as for the regional contraction of all forms of agri-
culture.

Conclusions

Data from the Moche Valley support the hypothesis that agrarian
collapse in the Cordillera Negra is closely tied to uplift and tilting of
probable fault blocks along the Pacific watershed. The hydrological
consequences of the uplift include water table subsidence and river
entrenchment. The latter in turn is associated with the breakdown of
large scale irrigation systems by loss of canal intake efficiency and
eventual stranding of intakes above entrenching stream ff ow in bedrock
canyons.

Within the Moche drainage, the collapse rate registers as roughly a
25% loss of arable land per millennium. The collapse pattern registers
as both a “horizontal” contraction of farming toward the river and a
“vertical” retreat downslope. As a result, progressively less land can
be irrigated. In a like manner, coastal farming supported by ground-
water has constricted riverward and seaward, and modern irrigation
has moved into areas formerly farmed using groundwater. In the high-
lands terrace-agriculture has apparently also contracted inward and
shifted downslope.

Collapse represents both an ongoing gradual process and one punc-
tuated by radical environmental change. Episodic uplifts such as that
suggested by raised coastal terraces and Holocene beach ridges must
have extremely negative agricultural consequences. Episodic El Nino
ffash floods also have had demonstrably negative consequences. Flood-
ing at ca. A.D. 1 100 not only destroyed extant canal systems, but also
radically changed river erosion and course patterns to such a point that
the reconstructed irrigation system could not be effectively brought
back into operation, leading to its replacement by new downstream
canals that were not capable of supplying all of the land area that was
formerly farmed. In turn, these replacement canals have developed
chronic intake problems, leading to atrophy at their distal ends.

If agrarian collapse has been correctly analyzed, then future gener-
ations are not well served by the present practice of obliviously con-
structing major reclamation works within the uninvestigated ruins of
larger past systems. These ruins both gauge and foreshadow ongoing
environmental processes that modern technology is largely unaware of,
but to which it is not immune.

1983

Moseley et al. — Peruvian Agrarian Collapse

325

Acknowledgments

The research reported upon here was supported by grants from the National Science
Foundation (BNS77-24901), the Frederick Henry Prince Trust, and Soiltest, Inc. Per-
mission to conduct excavations in Peru was granted by the Instituto Nacional de Cultura,
Lima. We would like to acknowledge the assistance of Genaro Barr and Miguel Cornejo
(of Trujillo); Alan Kolata; James Kus, Fred Nials, and other members of Field Museum’s
Programa Riego Antiguo; and James Richardson, Thomas Anderson, Jack Donahue,
and Harold Rollins (of the University of Pittsburgh).

Literature Cited

Barazangi, M., and B. L. Isacks. 1976. Spatial distribution of earthquakes and sub-
duction of the Nazca plate beneath South America. Geology, 4:686-692.

Bawden, G. L. 1 982. Galindo: a study in cultural transition during the Middle Horizon.
Pp. 285-320, in Chan Chan, Andean desert city, (M. E. Moseley and K. C. Day,
eds.), Univ. New Mexico Press, Albuquerque.

Beck, C. M. 1979. Ancient roads on the north coast of Peru. Unpublished Ph.D. dissert..
Dept. Anthropology, Univ. California, Berkeley, 251 pp.

Bird, J. B. 1951. South American radiocarbon dates. Mem. Soc. Amer. Archaeology,
8: 37-49.

Cardenas Martin, M. 1977-78. Obtencion de una cronologia del uso de los recursos
marinos en el Antiguo Peru. Arqueologia PUC, 19-20:3-26.

Cossio, A., and H. Jaen. 1967. Geologia de los Cuadrangulos de Puemape, Chocope,
Otuzco, Trujillo, Salaverry, y Santa. Bol. Servicio de Geologia y Mineria, Lima, 17:
1-141.

Day, K. C. 1974. Walk-in-wells and water management at Chan Chan, Peru. Pp. 182-
190, in The rise and fall of civilizations (J. Sabloff and C. Lamberg-Karlovsky, eds.),
Cummings Publ. Co., Menlo Park, California.

Donnan, C., and C. R. Ortloff. 1983. Site tilt at the Chotuna Complex. Paper
presented at the 23rd annual meeting of the Institute of Andean Studies, Berkeley,
CA, 9 January 1983.

Eling, H. H., Jr. 1981. Prehispanic irrigation patterns: monadnocks of the Pampa de
Mojucape, Jequetepeque Valley, Peru. Paper presented at the Fourth Andean Ar-
cheology Colloquium, University of Texas, Austin, April 1981.

Ericksen, G. E., G. Plafker, and J. F. Concha. 1970. Preliminary report on the
geological events associated with the May 31, 1970 earthquake. United States Geo-
logical Survey Circular, 639:1-25.

Hastings, C. M., and M. E. Moseley. 1975. The adobes of Huaca del Sol and Huaca
de la Luna. Amer. Antiquity, 40:196-203.

Kjeatinge, R. W. 1974. Chimu rural administrative centers in the Moche Valley, Peru.
World Archaeology, 6:66-82.

Keatinge, R. W., and K. C. Day. 1973. Socio-economic organization of the Moche
Valley, Peru, during the Chimu occupation of Chan Chan. J. Anthro. Res., 29:275-
295.

Kolata, A. L. 1982. Chronology and settlement growth at Chan Chan. Pp. 67-86, in
Chan Chan, Andean desert city (M. E. Moseley and K. C. Day, eds.), Univ. New
Mexico Press, Albuquerque.

Kosok, P. 1965. Life, land and water in ancient Peru. Long Island Univ. Press, New
York, 264 pp.

Kulm, L. D., W. j. Schweller, A. Molina-Cruz, and V. J. Rosato. 1976. Lithologic
evidence for convergence of the Nazca plate with the South American continent.
Pp. 795-801, in Initial reports of the Deep Sea Drilling Project, U.S. Government
Printing Office, vol. 34.

326

Annals of Carnegie Museum

VOL. 52

Kus, J. S. 1972. Selected aspects of irrigated agriculture in the Chimu heartland,
Peru. Unpublished Ph.D. dissert., Dept. Geography, Univ. California, Los Angeles,
247 pp.

. 1983. The Chicama-Moche Canal: failure or success? An alternate explanation

for an incomplete canal. Paper presented at the 23rd Annual Meeting of the Institute
of Andean Studies, Berkeley, California, 9 January 1983.

Lange, T. Q. 1971. Preliminary studies of selected field systems, Moche Valley, Peru.
Unpublished B.A. dissert.. Dept. Anthropology, Harvard Univ., 1 14 pp.

Lettau, H. H., and K. Lettau. 1978. Exploring the world’s driest climate. Kept.
Institute for Environmental Studies, Univ. Wisconsin, Madison, 101:1-264.

Minster, J. B., et al. 1974. Numerical modeling of instantaneous plate tectonics.
Royal Astronomical Soc. Geophysical J., 36:541-576.

Moseley, M. E. n.d. Patterns of settlement and preservation in the Viru and Moche
Valleys. In press.

Moseley, M. E., and E. E. Deeds. 1982. The land in front of Chan Chan: agrarian
expansion, reform, and collapse in the Moche Valley. Pp. 25-54, in Chan Chan,
Andean desert city (M. E. Moseley and K. C. Day, eds.), Univ. New Mexico Press,
Albuquerque.

Moseley, M. E., and R. A. Feldman. 1982. Living with crises: a relentless nature
stalked Chan Chan’s fortunes. Early Man, 4:10-13.

Moseley, M. E., R. A. Feldman, and C. R. Ortloff. 1981. Living with crises: human
perception of process and time. Pp. 231-267, in Biotic crises in ecological and
evolutionary time (M. Nitecki, ed.). Academic Press, New York.

Moseley, M. E., and A. L. Kolata. 1982. Chan Chan: cloistered city . . . the home
of god-kings. Early Man, 4:6-9.

Nials, F. L., E. E. Deeds, M. E. Moseley, S. G. Pozorski, T. G. Pozorski, and R. A.
Feldman. 1979. El Nino: The catastrophic flooding of coastal Peru. Bull. Field
Mus. Nat. Hist., 50(7):4-14 (Part I); 50(8):4-10 (Part II).

ONERN. 1973. Inventario, evaluacion y uso racional de los recursos naturales de la
Costa: Cuenca del Rio Moche. Oficina Nacional de Evaluacion de Recursos Natu-
rales, Lima, 1:1-493; 2:1-109.

Ortloff, C. R., M. E. Moseley, and R. A. Feldman. 1982. Hydraulic engineering
aspects of the Chimu Chicama-Moche Intervalley Canal. Amer. Antiquity, 47:572-
595.

. 1 983<2. The Chicama-Moche Intervalley Canal: social explanations and physical

paradigms. Amer. Antiquity 48:376-389.

. 19836. Hydraulic engineering and historical aspects of the preColumbian intra-
valley canal systems of the Moche Valley, Peru. J. Field Arch., in press.

Plafker, G., G. E. Ericksen, and J. F. Concha. 1971. Geological aspects of the May
31, 1970 Peru earthquake. Bull. Seismological Soc. Amer., 61:543-578.

Pozorski, S. G., and T. G. Pozorski. 1979. Alto Salaverry: A Peruvian coastal pre-
ceramic site. Ann. Carnegie Mus. 48:337-375.

. 1982. Technical report on Programa Riego Antiguo field work. Pp. 68-72, in

The dynamics of agrarian collapse in coastal Peru, a report to the National Science
Foundation on Grant BNS77-24901 (by M. E. Moseley et al.).

. n.d. Preliminary results of investigations in the Casma Valley, Peru. Unpub-
lished manuscript, Section of Man, Carnegie Museum of Natural History, Pittsburgh,
170 pp.

Pozorski, T. G. 1982. Early social stratification and subsistence systems: the Caballo
Muerto complex. Pp. 225-254, in Chan Chan, Andean desert city (M. E. Moseley
and K. C. Day, eds.), Univ. New Mexico Press, Albuquerque.

Richardson, J. B., III. 1981. Modeling the development of sedentary maritime econ-
omies on the coast of Peru: a preliminary statement. Ann. Carnegie Mus. 50:139-
150.

1983

Moseley et al. — Peruvian Agrarian Collapse

327

Rollins, H. B., J. B. Richardson III, and D. H. Sandweiss. n.d. New evidence and
implications of a major change in the structure of the East Pacific water mass about
5000 years B.P. Unpublished manuscript. Dept. Geology, Univ. Pittsburgh.

Sandweiss, D. H., H. B. Rollins, and J. B. Richardson III. 1981. The preceramic
occupation of the uplifted Santa Valley coast and evidence for a major El Nino.
Unpublished manuscript. Dept. Anthropology, Cornell Univ.

ScHAEDEL, R. P. 1951. The lost cities of Peru. Scientific American, 185(2): 18-23.

Shimada, I. 1981. The Batan Grande-La Leche archaeological project: the first two
seasons. J. Field Archaeology, 8:405-446.

SiLLiTOE, R. H. 1974. Tectonic segmentation of the Andes: implications for magmatism
and metallogeny. Nature, 250:542-545.

Topic, T. L. 1982. The early intermediate period and its legacy. Pp. 255-284, in Chan
Chan, Andean desert city (M. E. Moseley and K. C. Day, eds.), Univ. New Mexico
Press, Albuquerque.

USDA. 1955. Water, the yearbook of agriculture, edited by Alfred Stefferud. United
States Dept. Agriculture, Washington, D.C., 751 pp.

West, M. 1981. Agricultural resource use in an Andean coastal ecosystem. Human
Ecology, 9:47-78.

Wyss, M. 1978. Sea level changes before large earthquakes. Earthquake Information
Bull., U.S. Geol. Surv., 10:165-168.

Back issues of many Annals of Carnegie Museum articles are
available, and a few early complete volumes and parts are listed
at half price. Orders and inquiries should be addressed to: Pub-
lications Secretary, Carnegie Museum, 4400 Forbes Avenue,
Pittsburgh, Pa. 15213.

73

ISSN 0097-4463

AN NALS

0/ CARNEGIE MUSEUM

CARNEGIE MUSEUM OF NATURAL HISTORY

4400 FORBES AVENUE * PITTSBURGH, PENNSYLVANIA 15213

VOLUME 52 16 SEPTEMBER 1983 ARTICLE 14

RESULTS OF THE ALCOA FOUNDATION-SURINAME
EXPEDITIONS. VIL RECORDS OF MAMMALS
FROM CENTRAL AND SOUTHERN SURINAME

Stephen L. Williams

Collection Manager, Section of Mammals

Hugh H. Genoways

Curator, Section of Mammals

Jane A. Groen

Scientific Preparator, Section of Mammals

Abstract

The occurrence of three species of mammals previously unknown in Suriname is
documented. The new taxa recorded include Vampyrops aurarius, Vampyrops lineatus,
and Natalus tumidirostris. Additional information is provided on Centwnycteris max-
imiliana, Sigmomys alstoni, Zygodontomys brevicauda, and Cavia aperea from Suri-
name.

Introduction

In October and November of 1981, the Section of Mammals of
Carnegie Museum of Natural History conducted its fourth field season
in Suriname. This trip included collecting at one locality in central
Suriname and at two localities in the southern part of the country.
Material acquired at these localities makes noteworthy contributions
to knowledge of the mammals of Suriname.

Fieldwork was conducted at Tafelberg, the Sipaliwini airstrip, and
Oelemarie. Tafelberg is a high plateau (500-1 100 m) in central Suri-

Submitted 4 April 1983.

329

330

Annals of Carnegie Museum

VOL. 52

name, and is located within the Tafelberg Nature Reserve (Genoways
et ah, 1982; Schultz et ah, 1977). The Sipaliwini airstrip is located in
south-central Suriname on the Sipaliwini Nature Reserve, which eco-
logically is part of the Paru savannah of Brazil (Genoways et ah, 1982;
Schultz et ah, 1977). Oelemarie is located in the southeastern corner
of the country where dense lowland rainforest predominates.

At these localities, we obtained seven species of mammals either
unknown previously from the country, known by a few specimens, or
restricted geographically in Suriname. Three species represent new rec-
ords of bats for the country, bringing the total number of bat species
known in Suriname to 96 (Davis, 1966; Husson, 1978; Genoways and
Williams, 1979, 1980; Genoways et al., 198 1; Williams and Genoways,
1980^2, 1980/?). New distributional records also were documented for
one other species of bat and three species of rodents that previously
were known to be part of the mammal fauna of Suriname.

Methods and Materials

All specimens collected were prepared as standard museum skins accompanied by
skulls, or were preserved in 10% formalin. All specimens are deposited in the Section
of Mammals of Carnegie Museum of Natural History.

Forearm and cranial dimensions were taken by means of dial calipers accurate to 0. 1
mm. External measurements, recorded in millimeters, were taken in the field by the
collector or preparator. Forearm and cranial measurements were taken as described by
Genoways and Williams (1979).

Field weights in grams were taken with Pesola spring scales. The reproductive condition
of standard museum specimens was determined by gross dissection in the field, whereas
fluid-preserved specimens were dissected in the laboratory. Crown-rump length of fetuses
and testes lengths of males were recorded in millimeters.

Species Accounts

Centronycteris maximiliani maximiliani (Fischer)

Specimen examined {\). — Marov^we\ Oelemarie, 3°06'N, 54°3FW, 1.

Husson (1978) reported the only previously known specimen from
Suriname from near the Tibiti River in Saramacca District. Our spec-
imen was taken about 300 km southeast of that locality.

The specimen from Oelemarie was an adult male captured on 22
November. It weighed 5 g and had testes that measured 2 mm. The
specimen was taken in dense rainforest with moderately developed
understory. Other species of bats caught in the same area included
Ptewnotus parnellii, Phyllostomus elongatus, Tonatia silvicola, Lon-
chophylla thomasi, CarolUa perspicillata, Rhinophylla pumilio, and two
species of Artibeus.

Measurements of our specimen are as follows: length of forearm.

1983

Williams et al. — Records oe Suriname Mammals

331

43.7; greatest length of skull, 15.1, condylobasal length, 14.0; zygomatic
breadth, 9.1; mastoid breadth, 7.7; postorbital breadth, 3.0; length of
maxillary toothrow, 5.9; breadth across upper molars, 6.7.

Vampyrops aumrius Handley and Ferris

Specimens examined (3). — Saramacca: SE side of Arrowhead Basin, Augustus Creek,
Tafelberg, 600 m, 03°54'N, 56°10'W, 2; Geyskes Creek, Tafelberg, 700 m, 03°55'N,
56°10'W, 1.

These are the hrst localities of record for Vampyrops aurarius in
Suriname. The nearest records of the species have been reported from
Guiana Highlands of southeastern Venezuela (Handley and Ferris, 1972;
Handley, 1976). Our specimens were collected in the largest highland
area (600 and 700 m) of Suriname. All specimens were captured with
mist nets placed over streams running through primary highland rain-
forest. Among the three females collected between 3 1 October and 3
November only one, from Arrowhead Basin, evinced any reproductive
activity; this specimen was carrying a 45 mm (crown-rump) embryo
on 31 October. The other two specimens are younger individuals in
which the phalangeal epiphyses are just closing. The dorsal pelage of
these two specimens is darker, shorter, and somewhat hner than that
of the adult female.

We compared our specimens with Venezuelan material of this species
in the National Museum of Natural History. Our specimens agree with
these in details of external, cranial, and dental characters as described
by Handley and Ferris (1972). The specimens from Suriname fall within
the range of measurements of material from Venezuela as reported by
Swanepoel and Genoways (1979). Measurements of our specimens are
as follows: length of forearm, 52.3, 51.9, 52.7; greatest length of skull,
29.5, 29.1, 28.8; condylobasal length, 26.4, 25.9, 25.8; zygomatic
breadth, 17.1, 17.1, 16.7; mastoid breadth, 14.2, 13.9, 13.9; postorbital
breadth, 6.7, 6.6, 6.6; length of maxillary toothrow, 10.7, 10.8, 10.3;
breadth across upper molars, 12.5, 12.7, 12.1.

Vampyrops lineatus E. Geoffroy

Specimens examined {\). — 'Sdc¥JE.R\E-. Sipaliwini airstrip, 2°02'N, 56°07'W, 1.

This specimen documents the occurrence of Vampyrops lineatus in
Suriname and on the Guyanan Shield. The nearest localities of record
are in northeastern Brazil, Bolivia, Peru, and Colombia (Gonzalez and
Vallejo, 1980; Koopman, 1982; Mares et al., 1981; Sanborn, 1955).

The specimen is a female that was carrying a nearly full-term embryo
(crown-rump length, 36 mm) when it was captured on 14 November.
The bat was caught in a net placed on the bank of the Sipaliwini River.
Vegetation in the area varied from dense secondary-growth to gallery

332

Annals of Carnegie Museum

VOL. 52

forest. Other species of bats collected at the locality included Loncho-
phylla thomasi, CaroUia perspicillata, three species of Artibeus, Stur-
nira tildae, Uroderma bilobatum, Lasiurus ega, Molossops planirostris,
and Molossus molossus.

This specimen was compared with specimens of V. lineatus housed
in the National Museum of Natural History and Carnegie Museum of
Natural History. It compares favorably with material from Brazil and
Bolivia, and its measurements match closely the published measure-
ments of specimens from Brazil, Paraguay, and Uruguay (Gonzalez
and Vallejo, 1980; Swanepoel and Genoways, 1979). Measurements
of our specimen are as follows: length of forearm, 47.2; greatest length
of skull, 25.7; condylobasal length, 22.5; zygomatic breadth, 15.1; mas-
toid breadth, 12.5; postorbital breadth, 6.4; length of maxillary tooth-
row, 8.7; breadth across upper molars, 10.2.

Natalus tumidiwstris haymani Goodwin

Specimen examined (1).—Nickerie: Sipaliwini airstrip, 2°02'N, 56°07'W, 1.

This specimen represents the first known record of the family Na-
talidae in Suriname. The species has been reported from Curasao (as
the subspecies N. t. tumidiwstris), Colombia and Venezuela {N. t. con-
tinentis), and Trinidad {N. t. haymania).

The specimen is a male that weighed 6.2 g and had testes measuring
2 mm when it was captured on 16 November. It was netted along a
trail that passed through mature tropical rainforest. Other species of
bats collected with this specimen included Pteronotus parnellii, Mi-
crony cteris mi nut a, Phylloderma stenops, Phyllostomus elongatus, P.
hastatus, Tonatia brasiliense, T. carrikeri, CaroUia brevicauda, C. per-
spicillata, Rhinophylla purnilio, three species of Artibeus, Sturnira lil-
ium, S. tildae, Uroderma bilobatum, Vampyrops helleri, Thyroptera
tricolor, and Myotis nigricans.

The specimen has a rostrum that is noticeably swollen so it hides
the molar teeth when viewed from above. Also, the posterior margin
of the palate is deeply emarginate to the level of the middle of the
second molar. These characteristics clearly identify the specimen as N.
tumidirostris and distinguish it from TV. stramineus (Goodwin, 1959).
We tentatively assigned our specimen to the subspecies haymani from
Trinidad rather than to continentis, which has been used for all main-
land specimens in the past. The specimen is larger cranially than any
reported specimen of TV. t. continentis and agrees more closely with
known material of TV. t. haymani (Goodwin, 1959). The specimen also
has a second upper premolar with a longitudinal axis that is not in a
parallel plane with those of the first and third premolars. Goodwin
0959) regarded this dental feature as characteristic of N. t. haymani.

1983

Williams et al.~ Records of Suriname Mammals

333

Clearly, a more detailed analysis will be necessary to thoroughly un-
derstand geographic variation in this species. Measurements of our
specimen are as follows: length of forearm, 40.7; greatest length of
skull, 17.4; condylobasal length, 15.7; zygomatic breadth, 8.6; mastoid
breadth, 8.0; postorbital breadth, 3.3; length of maxillary toothrow,
7.1; breadth across upper molars, 5.5.

Sigmomys alstoni savannamm Thomas

Specimen examined {\).—S\\ckek\e-. Sipaliwini airstrip, 2°02'N, 56°07'W, 1.

Husson (1978) reported two specimens of Sigmomys alstoni from
the savannah habitat at the Sipaliwini airstrip and two damaged skulls
that were removed from the stomach of a white-tailed kite {Elanus
leucurus) taken at Zanderij, Para District. Although Sigmomys appar-
ently occurs in the coastal savannah and in the savannah at Sipaliwini,
these few earlier records together with our extensive trapppng efforts
during four different field seasons indicate that either this species is
uncommon in Suriname or, for some reason, difficult to obtain.

Our specimen is a near adult female that was trapped during the day
on 1 2 November. The habitat was a small clearing in the grass bordering
the airstrip. The specimen weighed 74 g and was lactating. Examination
of reproductive organs revealed two placental scars. Other species of
rodents taken in the same grassy habitat bordering the airstrip included
Holochilus braziliensis, Oryzomys delicatus, Zygodontomys brevicau-
da, and Cavia aperea.

External and cranial measurements of our specimen are as follows:
total length, 252; length of tail, 107; length of hind foot, 30; length of
ear, 21; greatest length of skull, 32.8; condylobasal length, 30.6; zy-
gomatic breadth, 19.6; interorbital constriction, 5.4; mastoid breadth,
4.0; length of nasals, 12.4; length of maxillary toothrow, 6.4; length of
diastema, 8.5.

Husson (1978) discussed the use of the generic name Sigmodon
instead of Sigmomys for this species. He applied the name Sigmodon
because of the work of Hershkovitz (1955, 1966) and Cabrera (1961),
but thought Sigmomys should be recognized as a valid subgenus. We
elected to use the generic name Sigmomys because in our opinion the
grooved upper incisors of this species are sufficiently important to
warrant separation of this taxon at the generic level.

Zygodontomys brevicauda microtinus (Thomas)

Specimens examined (8).— -Nickerie: Sipaliwini airstrip, 2°02'N, 56°07'W.

Husson (1978) examined a large series of Z. brevicauda from Suri-
name, but all of the specimens were obtained in the coastal savannah

334

Annals of Carnegie Museum

VOL. 52

region. The southernmost locality of his material was 90 km inland.
Our specimens indicate the geographical distribution is considerably
more extensive in Suriname than previously was assumed. The spec-
imens from the Sipaliwini airstrip were taken approximately 320 km
south of the southernmost locality reported by Husson (1978).

Between 13 November and 18 November, eight individuals were
collected in grassy savannah habitat bordering the airstrip. Other rodent
species collected in the same habitat are listed in the account for Sig-
mornys alstoni. One male and one female were immature as indicated
by the lack of toothwear and by weights of only 20 and 30 g, respec-
tively; the male had testes measuring 5, whereas the female evinced
no gross reproductive activity. Three adult males had testes measuring
8, 9, and 10, and weights of 48, 55, and 62 g, respectively. Two adult
females contained two and three embryoes which had crown-rump
measurements of 3 and 4 mm. A third female was lactating, but evinced
no other gross reproductive activity. Weights of the three adult females
were 37, 45, and 52 g.

External and cranial measurements of three adult males and three
adult females, respectively, are as follows: total length, 215, 230, 245,
234, 219, 226; length of tail, 86, 95, 100, 99, 88, 92; length of hind
foot, 25,26, 28, 26, 26, 25; length of ear, 20, 1 9, 20, 2 1 , 20, 1 9; greatest
length of skull, 29.5, 31.6, 32.4, 30.4, 30.5, 29.7; condylobasal length,
27.6, 29.6, 30.3, 28.4, 28.3, 27.9; zygomatic breadth, 15.6, 16.6, 16.6,

16.1, 16.1, 15.9; interorbital constriction, 5.0, 5.1, 5.0, 4.9, 4.9, 4.9;
mastoid breadth, 12.5, 13.0, 13.2, 13.2, 12.4, 12.8; length of nasals,

12.2, 13.3, 12.9, 12.5, 12.3, 12.1; length of maxillary toothrow, 4.3,
4.5, 4.6, 4.4, 4.7, 4.5; length of diastema, 8.0, 8.7, 8.9, 8.3, 8.0, 8.0.

The systematic relationship between the Zygodontomys from Sipali-
wini and the northern coast currently is not understood because these
savannah habitats are separated by approximately 300 km of contin-
uous tropical rainforest. The name Z. brevicaiida microtinus is used
for these specimens primarily because this is the name applied by
Husson (1978) to populations in Suriname. Investigations currently
are being conducted on Zygodontomys in Suriname to determine their
taxonomic relationship.

Cavia aperea guianae Thomas

Specimens examined (3). — Nickerie: Sipaliwini airstrip, 2°02'N, 56°07'W, 3.

The existence of Cavia aperea in Suriname previously was based on
one specimen obtained by Husson (1978) near the Sipaliwini airstrip.
We obtained three additional specimens at the same locality. Clearly,
the species is rare in Suriname and probably is restricted to savannah
areas in the extreme southern portion of the country.

1983

Williams et al. — Records of Suriname Mammals

335

The three specimens are females, one of which is an adult and the
other two are juveniles. The adult was shot early in the morning of 14
November as it traveled with another adult and three young. The group
of Cavia had just left a tall grassy area and easily were seen on the
mowed grass surrounding the airport buildings. The adult female
weighed 380 g and was lactating. Examination of the uterus revealed
one placental scar. One juvenile was trapped on the same date; another
juvenile was trapped on 16 November. Both juveniles were trapped
during day time in tall grass bordering the airstrip. Weights for these
specimens were 83 and 1 10 g. They evinced no reproductive activity.

Observations from our collecting and information acquired from
local hunters indicate Cavia is relatively abundant at Sipaliwini. They
apparently are crepuscular to diurnal and may travel in small groups.
Distinct paths and cuttings were observed in many areas where the
grass was about 1 m in height. Local hunters commented that Cavia
are seen less frequently during dry periods. Other rodent species col-
lected in the same habitat are listed in the account for Sigmomys
alstoni.

External and cranial measurements of the adult female are as follows:
total length, 265; length of tail, 7; length of hind foot, 45; length of ear,
22; greatest length of skull, 59.3; condylobasal length, 54.6; zygomatic
breadth, 31.6; interorbital constriction, 11.7; mastoid breadth, 20.8;
length of nasals, 19.4; length of maxillary toothrow, 13.6; length of
diastema, 16.0.

We follow Husson (1978) in applying the name aperea to Cavia from
Sipaliwini. Husson (1978) briefly discussed the taxonomy of the species
in Suriname and gives his reasons for the use of this name.

Acknowledgments

Our fieldwork was supported by a grant from the Alcoa Foundation, Charles L. Gris-
wold, President. We gratefully acknowledge this support.

We would like to thank Mr. Henry A. Reichart, STINASU, for his assistance during
our work and for making the many facilities of STINASU available to us. Without his
help, our work in Suriname would have been impossible. Ferdinand L. J. Baal, De-
partment of Forestry, issued our permits, and Mr. Leo Roberts, STINASU, proved to
be an excellent field guide and a most congenial companion. Mr. Michael Arnold and
Mr. Ben Koop of Texas Tech University, Dr. Carleton J. Phillips and Mr. Keith Stud-
holme of Hofstra University, and Mr. Kris Mohadin and Ms. Muriel Held of STINASU
assisted with collection and preparation of specimens. Our gratitude is extended to Dr.
Alfred L. Gardner and Dr. Charles O. Handley, Jr., of the U.S. National Museum, for
their assistance in verifying certain species identifications.

Literature Cited

Cabrera, A. 1961. Catalogo de los mamiferos de America del Sur. II. (Sirenia-Peris-
sodactyla-Artiodactyla-Lagomorpha-Rodentia-Cetacea). Rev. Mus. Argentino Cicn.
Nat. “Bernardino Rivadavia,” Cien. Zook, 4:xxii + 309-732.

336

Annals of Carnegie Museum

VOL. 52

Davis, W. B. 1966. Review of South American bats of the genus Eptesicus. South-
western Nat., 11:245-274.

Genoways, H. H., H. a. Reichart, and S. L. Williams. 1982. The Suriname small
mammal survey: A case study of the cooperation between research and national
conservation needs. Pp. 491-504, in Mammalian biology in South America (M. A.
Mares and H. H. Genoways, eds.), Pymatuning Lab. Ecol. Univ. Pittsburgh, Spec.
Publ., 6:xii + 1-539.

Genoways, H. H., and S. L. Williams. 1 979. Records of bats (Mammalia: Chiroptera)
from Suriname. Ann. Carnegie Mus., 48:323-335.

. 1 980. Results of the Alcoa Foundation-Suriname Expeditions. I. A new species

of bat of the genus Tonatia (Mammalia: Phyllostomatidae). Ann. Carnegie Mus.,
49:203-211.

Genoways, H. H., S. L. Williams, and J. A. Groen. 1981. Results of the Alcoa
Foundation-Surname Expeditions. V. Noteworthy records of Surinamese mammals.
Ann. Carnegie Mus., 50:319-332.

Gonzalez, J. C., and S. Vallejo. 1980. Notas sobre Vampyrops lineatus (Geoffroy)
del Uruguay. Comun. Zool., Mus. Hist. Nat. Montevideo, 10:1-8.

Goodwin, G. G. 1959. Bats of the subgenus Natalus. Amer. Mus. Novitates, 1977:
1-22.

Handley, C. O., Jr. 1976. Mammals of the Smithsonian Venezuelan Project. Sci. Bull.
Brigham Young Univ., Biol. Ser., 20:1-91.

Handley, C. O., Jr., and K. C. Ferris. 1972. Descriptions of new bats of the genus
Vampyrops. Proc. Biol. Soc. Washington, 84:519-524.

Hershkovitz, P. 1955. South American marsh rats, genus Holochilus, with a summary
of sigmodont rodents. Fieldiana:Zool., 37:639-673.

. 1966. Mice, land bridges and Latin American faunal interchange. Pp. 725-

751, in Ectoparasites of Panama (R. L. Wenzel and V. J. Tipton, eds.). Field Mus.
Nat. Hist., Chicago, xii + 861 pp.

Husson, a. M. 1978. The mammals of Suriname. Zool. Monogr., Rijksmuseum Nat.
Hist., 2:xxiv + 1-569.

Koopman, K. F. 1982. Biogeography of the bats of South America. Pp. 273-302, in
Mammalian biology in South America (M. A. Mares and H. H. Genoways, eds.),
Pymatuning Lab. Ecol. Univ. Pittsburgh, Spec. Publ., 6:xii + 1-539.

Mares, M. A., M. R. Willig, K. E. Streilein, and T. E. Lacher, Jr. 1981. The
mammals of northeastern Brazil: A preliminary assessment. Ann. Carnegie Mus.,
50:81-137.

Sanborn, C. C. 1955. Remarks on the bats of the genus Vampyrops. Fieldiana:Zool.,
37:403-413.

Schulz, J. P., R. A. Mittermeier, and H. A. Reichart. 1977. Wildlife in Suriname.
Oryx, 14:133-144.

SwANEPOEL, P., AND H. H. Genoways. 1979. Morphometries. Pp. 13-106, in Biology
of bats of the New World family Phyllostomatidae (R. J. Baker, J. K. Jones, Jr., D.
C. Carter, eds.). Spec. Publ. Mus., Texas Tech Univ., 16:1-441.

Williams, S. L., and H. H. Genoways. 1980<3. Results of the Alcoa Foundation-
Suriname Expeditions. 11. Additional records of bats (Mammalia: Chiroptera) from
Suriname. Ann. Carnegie Mus., 49:213-236.

. 19807>. Results of the Alcoa Foundation-Suriname Expeditions. IV. A new

species of bat of the genus Molossops (Mammalia: Molossidae). Ann. Carnegie Mus.,
49:487-498.

7.3

ISSN 0097-4463

I

o/CARiNEClE MUSEUM

CARNEGIE MUSEUM OF NATURAL HISTORY

4400 FORBES AVENUE ^ PITTSBURGH, PENNSYLVANIA 15213

ARTICLE 15

16 SEPTEMBER 1983

VOLUME 52

TAXONOMIC REVIEW OF TEMMINCK’S TRIDENT BAT,
ASELLISCVS ri?/CE/YP/DAri7V(TEMMINCK, 1834)
(MAMMALIA: HIPPOSIDERIDAE)

Duane A. Schlitter

Associate Curator, Section of Mammals

Stephen L. Williams

Collection Manager, Section of Mammals

John Edwards Hill^

Abstract

Specimens of Temminck’s trident bat {Aselliscus tricuspidatus) from Australasia were
analyzed for morphological variation. Univariate and multivariate analyses were used
to determine individual, secondar>' sexual, and geographic variation within the species.
Individual variation was greater in females than in males although this variation generally
was minor in all cases. Females were found to be significantly larger than males in four
of seven measurements tested. The analysis of geographic variation revealed that four
subspecies can be recognized within the species; two of these are described as new.

Introduction

Two species of trident bats are currently recognized in the genus
Aselliscus (Corbet and Hill, 1980:52). Temminck’s trident bat, Asel-
liscus tricuspidatus (Temminck, 1834), is known from Buru and Batjan
islands in the Moluccas eastward through New Guinea and the Solo-
mons as far as Espiritu Santo Island in the New Hebrides (Laurie and
Hill, 1954:62). The second species in the genus, Aselliscus stoliczkanus
(Dobson, 1871), occurs from Burma and southern China southward

* Address: Department of Zoology, British Museum (Natural History), London, England
SW7 5BD.

Submitted 14 April 1983.

337

338

Annals of Carnegie Museum

VOL. 52

through southeastern Asia as far as Penang Island in peninsular Ma-
laysia.

Aselliscus tricuspidatiis (Temminck, 1834) was proposed for a female
specimen collected by Muller and Macklot from Ambon Island, Mo-
luccas. A subspecies, A. t. novehebridensis, was described by Sanborn
and Nicholson (1950:331) based on specimens from Espiritu Santo
Island, New Hebrides, at the eastern limits of the range of the species.

Although few specimens are available from some areas within the
known range of the species, enough specimens are available at this time
to review the taxonomy of the species. The objectives of the study were
to analyze nongeographic and geographic variation within the species,
and then to assess the taxonomic significance of the variation.

Materials and Methods

A total of 2 1 1 specimens of Aselliscus tricuspidatiis were examined. Only adults were
used for statistical comparisons; these were so judged by the complete fusion of the
epiphyseal phalanges and of the sutures in the basioccipital-basisphenoidal region of the
skull. Conventional standard external measurements, when given, were recorded from
the specimen label. Forearm and cranial measurements were taken with dial calipers.

One external and six cranial measurements were selected for comparison and are given
in millimeters. Measurements are defined below.

Length of forearm. — Gvcdiitsl distance from posterior extremity of the olecranon pro-
cess of the ulna to the distal extremity of the carpals.

Condylocanine — Greatest distance from posterior portion of occipital condyles

to anteriormost edge of canines.

Zygomatic breadth. — GrcdiXQsX width across zygomatic arches, measured at right angle
to longitudinal axis of cranium.

Post orbital constriction. — Lq2lsX width across constriction posterior to orbits, measured
at right angle to longitudinal axis of cranium.

Mastoidal breadth. — GvcdiX^sX width across mastoidal processes, measured at right
angle to longitudinal axis of cranium.

Length of maxillary roo?/? row. — Least distance from anterior lip of alveolus of C to
posterior lip of alveolus of ML

Breadth across upper molars. — LqslsX distance, measured at right angle to longitudinal
axis of cranium, from labial side of respective crowns of M^ of each toothrow.

Localities from which specimens were examined were grouped into 13 geographical
samples for analyses of data. These samples (identified by number, see Fig. 1), with the
localities included in each group, are as follows: 1) Indonesia: Buru; 2) Indonesia: Ambon,
Seram, and Gorong Islands; 3) Indonesia: Kai Islands; 4) Indonesia: Misor Islands and
Yapen Island; 5) Indonesia: Irian Jaya: Hollandia and south side Humboldt Bay; 6)
Papua New Guinea: Kaiseren Augusta River; Ama; 7) Papua New Guinea: Madang; 8)
Papua New Guinea: Woodlark Island; Misima Island; and Kiriwina Island; 9) Solomon
Islands: Russell Islands; New Georgia Island; San Jorge Island; and Guadalcanal; 10)
Solomon Islands: Rennell Island; 1 1) Solomon Islands: San Cristobal Island and Uki Ni
Masi; 12) Santa Cruz Island; 13) New Hebrides: Espiritu Santo Island.

Univariate analyses of secondary sexual variation and individual variation were per-
formed using the UNI VAR program developed and introduced by Power (1970). Stan-
dard statistics (mean, range, standard deviation, standard error, variance, and coefficient
of variation) are generated by this program. A single classification analysis of variance
(ANOVA; F-test, significance level 0.05) was employed to test for significant differences

1983

SCHLITTER ET \L. — ASELLISCUS TAXONOMY

339

o

o

CM

O

r-

o

CD

o

o

o

CO

Fig. l.~ Geographic areas included in the 13 geographic samples of Aselliscus tricuspidatus. See Materials and Methods section of text
for list of localities included in each sample.

340

Annals of Carnegie Museum

VOL. 52

between or among means (Sokal and Rohlf, 1969). When means were found to be
significantly different, the Sums of Squares Simultaneous Test Procedure (SS-STP) de-
termined the maximally nonsignificant subsets.

Stepwise discriminant analysis and canonical analysis (BMDP7M, Dixon and Brown,
1977) perform a multiple discriminant analysis in a stepwise fashion, selecting the vari-
able entered by finding the variable with the greatest F value. The F value for inclusion
was set at 0.01, and the F value for deletion was set at 0.05. Canonical coefficients were
derived by multiplying the coefficient of each discriminant function by the mean of each
corresponding variable. The program also classifies individuals by placing them with the
group that they are nearest to on the discriminant functions.

Results

Nongeographic Variation

Using a single classification analysis of variance, individual and sec-
ondary sexual variations were examined in a single sample of Aselliscus
tricuspidatus from the vicinity of Hollandia, Dutch New Guinea [=Jay-
apura, Irian Jaya, Indonesia] (Table 1). Age variation was not consid-
ered because only adults were available for study.

Individual — Coefficients of variation (CV) of one external

and six cranial measurements are represented in Table 1. CVs ranged
from 1.05 (mastoidal breadth of females) to 5.45 (postorbital constric-
tion of females). Both CVs for postorbital constriction were above 5
(5.19 in males) whereas the next highest CV was 2.47 for length of
maxillary toothrow of females.

Coefficients of variation are listed in Table 2 for additional geo-
graphic samples of males and females. These values generally match
the low values listed in Table 1. Among the geographic samples of
males, only postorbital constriction shows a high CV value. In the
geographic samples of females, again the postorbital constriction ex-
hibits high CV values. In addition, OTU 2 has high CV values for
length of forearm and breadth across upper molars while OTU 4 has
a high CV value for mastoidal breadth and OTU 9 has a high value
for length of maxillary toothrow. Other than postorbital constriction,
which was the smallest measurement taken, coefficients of variation
for all of the measurements in this study were generally low and in
accordance with similar measurements of previous studies of bats (Long,
1968; Hayward, 1970; Smith, 1972; Davis, 1973; Swanepoel and Gen-
oways, 1979; Martin and Schmidly, 1982).

Secondary sexual variation. — Based upon the results of a single clas-
sification analysis of variance between males and females from Hol-
landia, Irian Jaya, females were found to be significantly larger than
males in four of the seven measurements tested for secondary sexual
variation (Table 1). The length of maxillary toothrow showed the high-
est value for significant differences between the two sexes. In two of
the remaining three measurements, females averaged larger than males.

1983

SCHLITTER ET AL.~ ASELLISCUS TAXONOMY

341

Table L— Nongeographic variation in selected externa! and cranial measurements of 15
females and 12 males of Aselliscus tricuspidatus from the vicinity of Jayapura, Irian
Jaya, Indonesia f=Hoilandia, Dutch New Guinea]. Statistics given are mean, two stan-
dard errors of the mean, range, coefficient of variation, F-value, and critical F-value (lower
value). Means that were found to be not significantly different at the 5 per cent level are

marked ns.

Sex

Mean ± 2 SE

Range

cv

Fs/F

Length of forearm

Female

42.19 ± 0.42

41.0-43.3

1.93

6.85

Male

41.29 ± 0.56

39.2-42.6

2.33

4.24

Condylocanine length

Female

13.19 ± 0.08

13.0-13.4

i.il

4.57

Male

13.04 ± 0.12

12.8-13.5

1.58

4.24

Zygomatic breadth

Female

7.63 ± 0.07

7. 3-7. 8

1.68

ns

Male

7.56 ± 0.08

7.3-7. 8

1.74

Postorbital constriction

Female

1.95 ± 0.05

1. 7-2.1

5.45

ns

Male

1.98 ± 0.06

1. 8-2.2

5.19

Mastoidal breadth

Female

7.04 ± 0.04

6. 9-7.2

1.05

ns

Male

6.98 ± 0.07

6. 8-7.2

1.63

Length of maxillary toothrow

Female

5.05 ± 0.06

4.8-5. 2

2.47

15.20

Male

4.89 ± 0.05

4.8-5. 1

1.62

4.24

Breadth across upper molars

Female

5.30 ± 0.06

5.0-5. 4

2.26

6.76

Male

5.19 ± 0.05

5.0-5. 3

1.73

4.24

In no instance were males significantly larger than females. Thus be-
cause females are significantly larger than males in some measurements,
samples of males and females were kept separate for the purpose of
analyzing geographic variation.

Geographic Variation

Univariate and multivariate analyses were performed to compare
the geographic samples in order to establish the relationships among
the populations studied. Specimens of Aselliscus tricuspidatus were
grouped into 13 geographic samples (OTUs) separated by sexes, for
the analysis of geographic variation (for localities included in each
geographic sample, see Materials and Methods section and Fig. 1), No
females were available for geographic samples 12 and 13.

342

Annals of Carnegie Museum

VOL. 52

Univariate analyses. — GQOgxdiphic samples of three or more individ-
uals were entered in the Sum of Squares Simultaneous Test Procedure
(SS-STP) to determine the maximally nonsignihcant subsets between
and among means of each variable for the geographic samples. Table
2 gives standard statistics for the seven variables from the 1 3 geographic
samples. Table 3 lists a series of means with maximally nonsignihcant
subsets of the seven variables in order to demonstrate trends in geo-
graphic variation.

Results of the univariate analyses of geographic variation indicate
that the New Hebrides sample (OTU 1 3) is characterized by individuals
of large size for the species. Individuals from mainland New Guinea
(OTUs 4-7) are slightly smaller in most variables than those from New
Hebrides. Two geographical groups of OTUs are characterized by small
individuals, namely the samples from the Moluccas (OTUs 1-3) and
the East Papuan islands and the Solomons (OTUs 8-1 1). Individuals
from Santa Cruz (OTU 1 2) are small but tend in some measurements
to be as large as individuals from New Hebrides.

Patterns of geographic variation indicate that the New Hebrides
sample (OTU 13) is large for measurements of length of forearm, con-
dylocanine length, zygomatic breadth and breadth across upper molars.
The New Guinea mainland group (OTUs 4-7) is large for the mea-
surements of length of forearm, condylocanine length (especially in
females), zygomatic breadth, and breadth across upper molars. The
western group (OTUs 1-3 from the Moluccas) show small size for the
measurements of condylocanine length, postorbital constriction, mas-
toidal breadth, and breadth across molars. This group exhibits rela-
tively large size for length of maxillary toothrow in males. Except for
postorbital constriction, the samples from the East Papuan islands and
the Solomons (OTUs 8-11) exhibit small size in all measurements,
especially in condylocanine length, zygomatic breadth, mastoidal
breadth, and breadth across upper molars in both sexes.

Multivariate analyses.— C?inomcdi\ analyses provide a procedure for
graphically representing phenetic relationship among samples with the
morphometric characters weighted by variance-covariance analysis.
The two-dimensional plots of the male and female samples of Aselliscus
tricuspidatus are presented in Fig. 2. In the males, small individuals
are separated to the right on the first variate as two groups. The Mo-
luccan samples (OTUs 1-3) and the East Papuan island sample (OTU
8) and the Solomon Islands samples (OTUs 9 and 11) are grouped
together. In the center, the Santa Cruz Island sample (OTU 12) is
grouped with the samples from Misor and Yapen islands of northeast
New Guinea (OTU 4) and the mainland New Guinea individuals (OTUs
5 and 7). On the left are the large individuals from New Hebrides (OTU
13). The Sepik River sample (OTU 6) is separated on Variate II on

1983

SCHLITTER ET ASELUSCUS TAXONOMY

343

Table 2. — Geographic variation in selected external and cranial measurements of samples
of males followed by females of Aselliscus tricuspidatus. Statistics given are sample size,
mean, Wo standard errors of the mean, range, and coefficient of variation. Means only
are listed for samples of less than four individuals. See Fig. 1 and text for key to locality
numbers and definition of measurements. No data for zygomatic breadth of males are

available for locality 10.

Locaiity

N

Mean ± 2 SE

Range

cv

MALES

Length of forearm

1

2

39.8, 38.6

2

6

39.7 ± 0.78

38.8-41.2

2.40

3

2

38.1, 37.8

4

1

41.4

5

12

41.3 ± 0.56

39.2-42.6

2.33

6

1

40.4

7

4

41.3 ± 1.33

39.6-42.6

3.23

8

15

39.2 ± 0.34

37.4-40.1

1.66

9

2

39.3, 37.5

10

2

39.3, 37.6

11

11

39.4 ± 0.50

37.7-40.5

2.11

12

3

39.0, 39.1, 37.4

13

2

41.5, 40.3

Condylocanine length

1

3

12.3, 12.3, 12.7

2

6

12.6 ± 0.96

12.4-12.7

0.96

3

2

12.2, 12.1

4

1

13.3

5

12

13.0 ± 0.12

12.8-13.5

1.58

6

1

13.0

7

4

13.1 ± 0.17

12.9-13.3

1.31

8

16

12.4 ± 0.08

12.1-12.7

1.34

9

2

12.4, 12.3

10

2

12.4, 12.1

11

10

12.5 ± 0.12

12.2-12.7

1.46

12

2

13.2, 12.3

13

2

13.5, 13.2

Zygomatic breadth

1

2

7.0, 6.9

2

5

6.8 ± 0.14

6.6-7.0

2.22

3

2

7.1, 7.0

4

1

7.9

5

12

7.6 ± 0.08

7. 3-7.8

1.74

6

1

7.7

7

4

7.6 ± 0.17

7. 4-7. 8

2.24

8

6

7.1 ± 0.10

6. 9-7. 2

1.71

9

2

7.0, 6.8

11

10

7.1 ± 0.09

6. 8-7. 3

2.08

12

2

6.7, 7.6

13

2

8.0, 7.8

344

Annals of Carnegie Museum

VOL. 52

Table 2. — Continued.

Locality

N

Mean ± 2 SE

Range

cv

Postorbital constriction

1

3

1.7, 1.7, 1.7

2

6

1.6 ± 0.10

oo

lA

7.41

3

2

1.9, 2.0

4

1

2.2

5

12

2.0 ± 0.06

1. 8-2.2

5.19

6

1

1.8

7

4

1.9 ± 0.10

1. 8-2.0

4.97

8

17

1.9 ± 0.04

1. 8-2.1

3.92

9

2

1.9, 1.7

10

2

2.0, 2.0

11

10

2.0 ± 0.03

1. 9-2.0

2.70

12

3

1.5, 1.9, 1.8

13

2

1.7, 1.8

Mastoidal breadth

1

3

6.4, 6.4, 6.6

2

5

6.7 ± 0.19

6.4-6.9

3.11

3

2

6.6, 6.5

4

1

7.2

5

12

7.0 ± 0.07

6. 8-7. 2

1.63

6

1

6.9

7

4

7.0 ± 0.13

6.8-7. 1

1.80

8

16

6.6 ± 0.09

6. 1-6.8

2.75

9

2

6.7, 6.6

10

2

6.8, 6.3

11

10

6.6 ± 0.07

6. 4-6. 7

1.56

12

3

6.6, 7.0, 6.8

13

2

7.0, 6.9

Length of maxillary toothrow

1

3

4.9, 4.9, 5.0

2

6

5.0 ± 0.08

4.8-5. 1

2.08

3

2

4.8, 4.8

4

1

5.2

5

12

4.9 ± 0.05

4.8-5. 1

1.62

6

1

5.5

7

4

5.1 ± 0.00

5.1

0.00

8

17

4.8 ± 0.04

4.7-5.0

1.69

9

2

4.8, 4.8

10

2

4.9, 4.8

1 1

11

4.9 ± 0.06

4.7-5.0

1.92

12

2

5.0, 5.2

13

2

5.2, 5.2

Breadth across upper molars

1

2

4.9, 4.9

2

6

4.9 ± 0.10

4. 7-5.0

2.38

3

2

4.9, 4.9

4

1

5.4

1983

SCHLITTER ET AL. — ^.SE'LL/i'CL/^ TAXONOMY

345

Table 1. — Continued.

Locality

N

Mean ± 2 SE

Range

cv

5

12

5.2 ± 0.05

5.0-5. 3

1.73

6

1

5.3

7

4

5.3 ± 0.10

5. 2-5. 4

1.82

8

17

4.9 ± 0.05

4.7-5. 1

2.10

9

2

5.0, 4.8

10

2

5.0, 4.8

11

1 1

4.9 ± 0.08

4.7-5. 1

2.58

12

3

4.6, 5.4, 4.8

13

2

5.5, 5.5

FEMALES

Length of forearm

1

1

40.5

2

3

39.8 ± 2.26

38.2-42.0

4.91

3

1

39.2

4

3

41.6 ± 1.22

40.4-42.4

2.54

5

15

42.2 ± 0.42

41.0-43.3

1.93

6

2

42.6, 41.3

7

2

42.1, 42.3

8

7

39.8 ± 0.40

38.9-40.4

1.33

9

4

39.2 ± 0.80

38.5-40.1

2.03

10

2

38.4, 38.4

11

9

39.4 ± 0.54

38.7-40.9

2.05

Condylocanine length

1

1

12.8

2

3

12.5 ± 0.23

12.3-12.7

1.60

3

1

12.7

4

3

13.3 ± 0.40

12.9-13.5

2.60

5

15

13.2 ± 0.08

13.0-13.4

1.11

6

2

13.5, 13.5

7

2

13.0, 13.1

8

8

12.5 ± 0.08

12.4-12.7

0.85

9

4

12.5 ± 0.25

12.2-12.8

2.00

10

2

12.4, 12.1

11

9

12.4 ± 0.07

12.2-12.5

0.88

Zygomatic breadth

1

1

7.0

2

3

7.0 ± 0.13

6.9-7. 1

1.66

3

1

7.1

4

3

7.5 ± 0.24

7. 3-7. 7

2.76

5

15

7.6 ± 0.07

7. 3-7. 8

1.68

6

2

7.5, 8.0

7

2

7.6, 7.6

8

2

7.0, 7.1

9

4

7.0 ± 0.13

6.8-7. 1

1.80

10

1

7.0

11

9

7.0 ± 0.07

6. 9-7. 2

1.50

346

Annals of Carnegie Museum

VOL. 52

Table 1. — Continued.

Locality

N

Mean ± 2 SE

Range

cv

Postorbital constriction

1

1

1.9

2

3

1.7 ± 0.18

1.5-1. 8

9.17

3

1

2.1

4

3

1.8 ± 0.13

1.7-1. 9

6.30

5

15

2.0 ± 0.05

1.7-2. 1

5.45

6

2

2.0, 2.2

7

2

2.0, 1.9

8

8

1.9 ± 0.05

1. 8-2.0

3.35

9

4

1.8 ± 0.14

1. 7-2.0

7.86

10

2

2.0, 1.8

11

9

1.9 ± 0.06

1.8-2. 1

4.34

Mastoidal breadth

1

1

6.5

2

3

6.7 ± 0.07

6.6-6. 7

0.87

3

1

6.7

4

3

7.3 ± 0.48

6. 8-7. 6

5.73

5

15

7.0 ± 0.04

6.9-7.2

1.05

6

2

7.1, 7.3

7

2

6.9, 7.1

8

8

6.7 ± 0.06

6.5-6.8

1.38

9

4

6.6 ± 0.15

6.5-6.8

2.28

10

2

6.6, 6.5

11

9

6.5 ± 0.07

6. 5-6. 8

1.55

Length of maxillary toothrow

1

1

5.0

2

3

4.8 ± 0.07

4.8-4. 9

1.19

3

1

5.0

4

3

5.2 ± 0.07

5. 1-5.2

1.12

5

15

5.1 ± 0.06

4.8-5.2

2.47

6

2

5.1, 5.0

7

2

5.1, 5.1

8

8

4.8 ± 0.05

4.7-4.9

1.54

9

4

4.9 ± 0.22

4.6-5. 1

4.55

10

2

4.9, 4.7

11

9

4.8 ± 0.06

4.6-4.9

1.93

Breadth across upper molars

1

1

4.9

2

3

5.0 ± 0.23

4.8-5.2

4.00

3

1

4.8

4

3

5.3 ± 0.12

5.2-5.4

1.89

5

15

5.3 ± 0.06

5.0-5.4

2.26

6

2

5.3, 5.4

7

2

5.4, 5.3

8

8

4.9 ± 0.05

4.8-5.0

1.54

9

4

4.9 ± 0.05

4.8-4. 9

1.03

10

2

4.8, 4.7

11

8

4.8 ±0.11

4.6-5.0

3.34

1983

SCHLITTER ET \h. — ASELLISCUS TAXONOMY

347

Table 3.— Results of seven SS-STP analyses of geographic variation in samples of males
followed by females of Aselliscus tricuspidatus. Vertical lines to the right of each series
of means connect maximally nonsignificant subsets at the 0.05 level of significance. See
Fig. 1 and text for key to geographic samples included in each locality number and
definitions of measurements.

Males

Females

Locality

Mean

Results

SS-STP Locality

Mean

Results

SS-STP

Length of forearm

5

41.3

\ 5

42.2

1

7

41.3

i i 4

41.6

i 1

2

39.7

1 1 8

39.8

1 1

11

39.4

1 2

39.8

1 1

8

39.2

1 11

39.4

1

9

39.2

1

Condylocanine length

7

13.1

\ 4

13.3

1

5

13.0

1 5

13.2

1

2

12.6

1 8

12.5

1

11

12.5

1 9

12.5

1

8

12.4

1 2

12.5

1

11

12.4

1

Zygomatic breadth

7

7.6

1 5

7.6

1

5

7.6

i 4

7.5

1

11

7.1

i 11

7.0

1

8

7.1

1 i 9

7.0

I

2

6.8

1 2

7.0

1

Postorbital constriction

5

2.0

1 5

2.0

1

11

2.0

1 11

1.9

1

7

1.9

1 8

1.9

1

8

1.9

1 4

1.8

1 1

2

1.6

i 9

1.8

i i

2

1.7

1

Mastoidal breadth

7

7.0

1 4

7.3

1

5

7.0

1 5

7.0

1

2

6.7

1 2

6.7

1

11

6.6

! 8

6.7

1

8

6.6

1 9

6.6

1

11

6.5

1

Length of maxillary toothrow

7

5.1

1 4

5.2

1

2

5.0

1 ! 5

5.1

1

11

4.9

1 9

4.9

1 1

5

4.9

1 8

4.8

1

8

4.8

1 2

4.8

1

11

4.8

1

348

Annals of Carnegie Museum

VOL. 52

Table 3. — Continued.

Males

Females

Locality

Mean

Results

SS-STP Locality

Mean

Results

SS-STP

7

5.3

Breadth across upper molars
\ 4

5.3

i

5

5.2

1 5

5.3

1

11

4.9

1 2

5.0

1

2

4.9

1 8

4.9

1

8

4.9

1 9

4.9

1

11

4.8

1

the upper left from the rest of the mainland New Guinea samples in
the center of the plot. In the females, most of the separation can be
accounted for on the first variate. Two groups are present in the plot
of females. The mainland New Guinea samples, including the Misor
and Yapen islands sample, are represented by OTUs 4-7 on the left
of the plot. The right of the plot includes two geographic groups. The
Moluccan samples are represented by OTUs 1-3 although the three
geographic samples are separated somewhat by Variate II. The samples
from the East Papuan islands and the Solomon Islands are grouped on
the left on Variate I. OTU 9 is separated from the other samples on
Variate II.

The amount of total dispersion accounted for in the male and female
samples, respectively, was 50.8 and 79.3% for Variate I, and 36.7 and
11.5% for Variate II. The total dispersion accounted for on the first
two axes for male and female samples, respectively, was 87.5 and
90.8%.

Morphometric characters used in this analysis are listed in Table 4,
from the most useful to the least useful in discriminating groups. For
males, zygomatic breadth, mastoidal breadth, and breadth across upper
molars have the strongest influence in separating those samples on the
first variate of the discriminant projection plot of Fig. 2. The length of
maxillary toothrow separates OTU 6 from the remaining groups while
the length of postorbital constriction predominantly separates the groups
in the lower right of the projection plot. Condylocanine length separates
those groups in the lower left of the projection plot. In females, all
seven variables equally separate the two major groups on the first
variate while zygomatic breadth and postorbital constriction have a
positive effort on Variate II.

The classification matrix for males and females of Aselliscus tri-

1983

SCHLITTER ET Ah. — ASELLISCUS TAXONOMY

349

Fig. 2. — Two-dimensional projection of the first two canonical variates of male (upper)
and female (lower) geographic samples of Aselliscus tricuspidatus, based on a classification
of variance-covariance among one external and six cranial measurements.

350

Annals of Carnegie Museum

VOL. 52

Table 4. — Variables used in discriminant function analyses of males and females of
Aselliscus tricuspidatus. Characters are listed in order of their usefulness in distinguishing
groups, with the character with the greatest between- groups variance and the least within-
groups variance being selected first. The statistics are recalculated at each step.

Step

Character

F-value

U-statistic

1

Males

Zygomatic breadth

25.47

0.1054

2

Length of maxillary toothrow

13.06

0.0192

3

Postorbital constriction

6.76

0.0056

4

Mastoidal breadth

3.63

0.0024

5

Condylocanine length

2.13

0.0013

6

Length of forearm

1.41

0.0009

7

Breadth across upper molars

0.47

0.0007

1

Females

Condylocanine length

21.70

0.1250

2

Postorbital constriction

3.68

0.0562

3

Zygomatic breadth

2.70

0.0291

4

Length of forearm

1.86

0.0175

5

Mastoidal breadth

1.67

0.0108

6

Breadth across upper molars

0.87

0.0081

7

Length of maxillary toothrow

0.95

0.0059

cuspidatus analysed in the canonical analysis is given in Table 5. All
females were correctly classified for each geographical sample. For
males, two individuals from the vicinity of Hollandia and Humboldt
Bay were placed with the Sepik River sample while five were classified
with the Madang sample. A single male from the Solomon Islands
sample was placed with the Kai Island sample, both groups being
generally small in size. A single male from San Cristobal Island was
classified with the Rennell Island sample.

The geographic variation found in A. tricuspidatus is almost exclu-
sively metrical in nature, with variation occurring between different
islands and island groups. There exists overlap in dimensional features
between populations from widely separated islands or between islands
that differ sharply in size, in the latter case probably directly reflecting
the size of the population that the island can support. Although color
of pelage was checked in the various geographic samples studied, no
discernable geographic variation in pelage color was noted even though
individual variation from yellowish brown to orange and even reddish
color was recorded.

Taxonomic Conclusions

Based upon our interpretations of the univariate and multivariate
analyses, we found that Aselliscus tricuspidatus can be divided into four

1983

SCHLITTER ET \l.. — ASELLISCVS TAXONOMY

351

Table 5. — Classification matrix for male and female samples of Aselliscus tricuspidatus,
based upon the discriminant functions of seven morphometric characters. Values indicate
the number of individuals classified into each group. Sample 10 and samples 12 and 13
are unrepresented in the male and female classification matrices, respectively.

Sample

Classification

groups

1

2

3

4

5

6

7

8

9

10

1 1

12

13

Males

1) Bum

2

0

0

0

0

0

0

0

0

0

0

0

2) Anbon and Seram

0

5

0

0

0

0

0

0

0

0

0

0

3) Kai I.

0

0

2

0

0

0

0

0

0

0

0

0

4) Misor and Yapen

0

0

0

1

0

0

0

0

0

0

0

0

5) Humboldt Bay

0

0

0

0

12

0

0

0

0

0

0

0

6) Sepic River

0

0

0

0

0

1

0

0

0

0

0

0

7) Madang

0

0

0

0

0

0

4

0

0

0

0

0

8) East Papuan Is.

0

0

0

0

0

0

0

5

0

0

0

0

9) Solomon Is.

0

0

0

0

0

0

0

0

2

0

0

0

1 1) San Cristobal I.

0

0

0

0

0

0

0

0

0

8

0

0

1 2) Santa Cruz

0

0

0

0

0

0

0

0

0

0

1

0

1 3) New Hebrides

0

0

0

0

0

0

0

0

0

0

0

2

Females

1) Buru

1

0

0

0

0

0

0

0

0

0

0

2) Ambon and Seram

0

3

0

0

0

0

0

0

0

0

0

3) Kai I.

0

0

1

0

0

0

0

0

0

0

0

4) Misor and Yapen

0

0

0

3

0

0

0

0

0

0

0

5) Humboldt Bay

0

0

0

0

8

2

5

0

0

0

0

6) Sepic River

0

0

0

0

0

2

0

0

0

0

0

7) Madang

0

0

0

0

0

0

2

0

0

0

0

8) East Papuan Is.

0

0

0

0

0

0

0

2

0

0

0

9) Solomon Is.

0

1

0

0

0

0

0

0

3

0

0

10) Rennell I.

0

0

0

0

0

0

0

0

0

1

0

1 1) San Cristobal I.

0

0

0

0

0

0

0

0

0

1

7

distinct geographic groups based upon size. The largest size for the
species is found in individuals from New Hebrides. Slightly smaller in
size are individuals from the New Guinea mainland and Misor and
Yapen islands. Individuals from the Moluccas and from the islands
east of the mainland of New Guinea as far east as Santa Cruz Island
are among the smallest in size. Although the individuals from Santa
Cruz Island approach the New Hebrides sample in size in certain mea-
surements, they are nonetheless nearer to the populations immediately
to the west.

The samples from the Moluccas are referrable \o A. t. tricuspidatus;
those from New Hebrides to Aselliscus tricuspidatus novehebridensis
Sanborn and Nicholson, 1950. The remaining two geographic groups
are described as new subspecies in the following accounts.

352

Annals of Carnegie Museum

VOL. 52

Accounts of Subspecies

Aselliscus tricuspidatus tricuspidatus (Temminck, 1834)

1834. Rhinolophus tricuspidatus Temminck, Tijdschr. Natuurk. Gesch., 1(1):20, pi. 1,
fig. 4.

1871. PhvUorhina tricuspidata, Peters, Monats. K. Preuss. Akad. Wissensch., Berlin, p.
314.'

1904. Hipposiderus tricuspidata, Trouessart, Cat. Mamm., Suppl., p. 70.

1941. Aselliscus tricuspidatus, Tate, Amer. Mus. Nov., 1 140:2.

//o/c/ypp. — RMNH “cat. a, v. sp. a,” female; from Ambon Island,
Moluccas, Indonesia.

Measurements of /zo/ofvpc — Length of forearm, 39.3; condylocanine
length, 12.5; zygomatic breadth, 7.1; width of postorbital constriction,
1.7; mastoidal breadth, 6.7; alveolar length of maxillary toothrow, 4.9;
and breadth of palate at 5.2.

Distribution. — from the Molucean islands of Morotai, Batjan,
Buru, Ambon, Seram, Gorong, and Kai.

Diagnosis. — ExXQvndiWy and cranially small-sized for the species (Ta-
bles 2, 3); forearm averaging small; cranially small, interorbital and
mastoidal region narrow, breadth of palate narrow, and skull short;
maxillary toothrow long relative to skull length in males.

Specimens examined (29). — Indonesia: Moluccas, Buru Island, Leksoela, 4 (4 BMNH);
Batjan Island, 6 (6 BMNH); Ambon, 1 (1 RMNH, holotype); Seram, 14 (7 BMNH, 7
ZMA); Gorong Islands, 1 (1 BMNH); Kai Islands, 3 (3 BMNH).

Additional records. — Indonesia.: Moluccas, Buru Island, Mefa; Leksula (Dammerman,
1929:161). Indonesia: Moluccas, Kai Islands (Peters and Doria, 1880:692).

Remarks. — Dobson ( 1 878c: 1 32) reports Aselliscus tricuspidatus from
Morty Island [=Morotai Island] from the Halmahera group in the
northern Molucea Islands. Dobson’s specimen is a spirit-preserved,
unregistered specimen in a poor state of preservation in the collection
of the British Museum (Natural History). Such measurements as could
be taken are as follows: length of forearm, 38.5; zygomatic width, 6.8;
postorbital constriction, 1.7; length of the maxillary toothrow, 5.1; and
breadth of palate at M^-M^, 4.9. The sex of the specimen could not
be established. It is referred to the nominate subspecies.

Aselliscus tricuspidatus novehebridensis Sanborn and Nicholson, 1950

1950. Aselliscus tricuspidatus novehebridensis Sanborn and Nicholson, Fieldiana Zook,
31:331.

Holotype. — YMNYX 55219, male; from cave on Segond Channel,
Espiritu Santo Island, New Hebrides.

Measurements of holotype.— Total length, 60; length of tail, 20; length
ofhindfoot, 7; length of ear, 12; length of forearm, 41.5; condylocanine
length, 13.5; zygomatic breadth, 8.0; width of postorbital constriction.

1983

SCHLITTER ET M.. — ASELLISCVS TAXONOMY

353

1.7; mastoidal breadth, 7.0; alveolar length of maxillary toothrow, 5.2;
and breadth of palate at 5.5.

Distribution. — YsSiov^n from the islands of Espiritu Santo, Aore, and
Malekula.

Diagnosis.— ExXQrmWy and cranially large-sized for species; length
of forearm long; condylocanine length of skull long; palate and zygoma
broad.

Specimens examined (2). — Espiritu Santo Island, New Hebrides, 2 (2 FMNH, including
holotype).

Additional records. — Hebrides: Aore Island, Aouta; Aore Island, Aouta Planta-
tion; Espiritu Santo Island, Hog Harbor; and Malekula Island, Senwar Cave, Tenmial,
northwest coast of island (Hill, 1983:151).

Remarks. — WiW (1983:151-152) reports 34 additional specimens of
this subspecies from three islands in the New Hebrides and also gives
measurements. Based upon this new material, it is evident that A. t.
novehebridensis is larger than was indicated by the description, with
forearm length reaching 43.5 in females and 42.2 in males in this new
material.

AselUscus tricuspidalus koopmank new subspecies

Holotype. — Adult female, skin and skull, AMNH 159400; Liluta, 10
m, Kiriwina Island, Papua New Guinea, obtained on 15 December
1956 by R. F. Peterson, original number 14,977, on the Fifth Archbold
Expedition to New Guinea.

Measurements of holotype. — Total length, 58; length of tail, 19; length
of hindfoot, 7; length of forearm, 40.4; condylocanine length, 12.4;
zygomatic breadth, 7.1; width of postorbital constriction, 2.0; mastoidal
breadth, 6.7; alveolar length of maxillary toothrow, 4.9; and breadth
of palate at 4.9.

Distribution. — ¥sAio\^n from New Ireland and New Britain islands in
the Bismarck Archipelago, Louisiade Archipelago, Trobriand Islands,
Woodlark Island, and Solomon Islands to Santa Cruz Island.

Diagnosis. — Externally and cranially small for species; length of fore-
arm short; skull short and narrow, with relatively broad interorbital
region relative to skull size.

Comparisons.— AselUscus tricuspidatus koopmani can be distin-
guished from the adjacent subspecies of the species (see Fig. 3) by its
smaller size, both externally and cranially.

Etymology.— This new subspecies is named for Karl F. Koopman, Curator of Mam-
mals, American Museum of Natural History, New York, in recognition of his keen
interest in Australasian bats and for first mentioning (Koopman, 1982:17) the small size
of specimens included in this new subspecies.

Specimens examined {\7>%). — TdLp\x2L New Guinea: New Ireland, no specific locality, 1
(1 BMNH); Papua New Guinea: Woodlark Island, Kulumadau, 200 m, 52 (52 AMNH);

354

Annals of Carnegie Museum

VOL. 52

northern Molucca Islands.

1983

SCHLITTER ET W.. — ASELLISCUS TAXONOMY

355

Papua New Guinea: Woodlark Island, no specific locality, 1 (I BMNH); Papua New
Guinea: Louisiade Archipelago, Misima Island, Kulumalia Mine, 1 50 m, 32 (32 AMNH);
Papua New Guinea; Trobriand Islands, Kiriwina Island, Liluta, 3 (3 AMNH, including
holotype); Solomon Islands: New Georgia Island, Munda Point, 8 (8 FMNH); Solomon
Islands: New Georgia Island, no specific locality, 1 (1 BMNH); Solomon Islands: San
Jorge Island, Talise, 1 (1 BMNH); Solomon Islands: Russell Islands, Banika Island, 1 (1
FMNH); Solomon Islands: Malaita Island, Riba Caves, near King George V School,
Auki, 1 (1 BMNH); Solomon Islands: Guadalcanal, Aola, 2 (2 BMNH); Solomon Islands:
Rennell Island, Tigoa, 4 (4 BMNH); Solomon Islands: San Cristobal, Warihito River—
Goge River confluence, about 6 mi inland from Wainoni Bay, 27 (27 BMNH); Solomon
Islands: Uki Ni Masi, 1 (1 FMNH); Santa Cruz Islands: Santa Cruz Island, 3 (3 AMNH).

Additional records. — Solomon Islands: Rennell Island, Kogoata Cave, Tigoa; Niupani
(Hill, 1968:56). Papua New Guinea: New Ireland Province, near Sohun (Smith and Hood,
1981:108). Papua New Guinea: East New Britain Province, near Gunanur Plantation
(Smith and Hood, 1981:108). Papua New Guinea: East New Britain Province, Duke of
York Island (Dobson, 1877:121; 1 878^2:3 17).

Remarks. — ThQ specimens from Santa Cruz Island and Uki Ni Masi
[=Ugi Island] were originally reported by Sanborn (1931:24) as Hip-
posiderus tricuspidatus.

Specimens from New Ireland and New Britain islands are assigned
to this new subspecies based upon the individual examined from New
Ireland and the measurements reported by Smith and Hood (1981:
107). Variation in larger samples from these two major islands should
be checked further to verify the subspecies to which they actually be-
long.

Aselliscus tricuspidatus novaeguineaCy new subspecies

Holotype. — KdixAx female, skin and skull, CM 63498; Ama, 140 m.
East Sepik Province, Papua New Guinea (04°09'S, 141°4rE), obtained
on 9 March 1980 by Stephen L. Williams, original number 5089,
Bernice P. Bishop— New Guinea held series BBM-NG 107652.

Measurements of holotype. — Total length, 71; length of tail, 25; length
ofhindfoot, 8; length of ear, 15; length of forearm, 41.3; condylocanine
length, 13.5; zygomatic breadth, 8.0; width of postorbital constriction,
2.2; mastoidal breadth, 7.3; alveolar length of maxillary toothrow, 5.0;
and breadth of palate at M^-M^, 5.4.

Distribution. — Now Guinea and adjacent islands of Misor and Yapen
in Irian Jaya.

Diagnosis. — Extornally and cranially medium- to large-sized for
species; length of forearm long; skull long and broad, condylocanine
length long, and palate and zygoma broad.

Etymology. — This new subspecies is named for the island of New Guinea.

Specimens examined (42). — Irian Jaya, Schouten Islands, Misor Island, Korido, 1 (1
BMNH); Irian Jaya, Geelvink Bay District, Yapen Island, 1 mi NW Sumberbaba, 1000
ft, 2 (2 AMNH); Irian Jaya, Yapen Island, Dawai River, 1 (1 AMNH); Irian Jaya,
Jayapura, 25 (20 AMNH, 3 USNM, 1 MVZ, 1 FMNH); Irian Jaya, South side Humboldt
Bay, 4 (4 RMNH); Papua New Guinea: West Sepik Province, Kaiseren Augusta River

356

Annals of Carnegie Museum

VOL. 52

(4°4'18"S, 141°7'15"E), 1 (1 RMNH); Papua New Guinea: East Sepik Province, Ama,
140 m, 2 (1 CM, 1 PNGM, including holotype and BBM-NG 105921 to be deposited
in PNGM); Papua New Guinea: Madang Province, Sempi, 13 mi N Madang, 4 (4 MVZ);
Papua New Guinea: Madang Province, Sempi Cave, 13.5 mi N, 1.5 mi W Madang, 1
(1 MVZ); Papua New Guinea: Madang Province, South Banup Cave, 6.5 mi S, 4.5 mi
W Madang, 1 (1 MVZ).

Additional records.— Guinea: no specific locality (Dobson, 1878/):876). Papua
New Guinea: Madang Province, 8 km E Baku, Gogol Valley, (Hill, 1983:152). Papua
New Guinea: Morobe Province, Huon Peninsula, no specific locality (Koopman, 1982:
16, [=Finschhafen, Koopman, in litl\). Papua New Guinea: Gulf Province, Putei (McKean,
1972:27). Papua New Guinea: East Sepik Province, Wagu (McKean, 1 972:27). Indonesia:
Irian Jaya, Andai (Peters and Doria, 1880:692).

Remarks. — two females from Ama were mist-netted in forest on
9 March 1980. Neither contained embryos although the holotype was
lactating. Each weighed 4 g. A male (MVZ 138607) from Sempi Cave,
Papua New Guinea, had testes measuring 3.5 mm in length and 1.5
mm in width when captured on 1 August 1 969. A female (MVZ 1 38605)
from Sempi had no embryos when checked on 24 July 1969. A male
(MVZ 138603) from South Banup Cave, 6.5 mi S, 4.5 mi W Madang,
Papua New Guinea, had testes measuring 3 by 2 mm when captured
on 25 August 1969. The record from Putei reported by McKean (1972:
27) is the only report of this species from the southern portion of New
Guinea. Additional collecting by mist-netting as well as visiting caves
should result in specimens from elsewhere in the region, especially in
Irian Jaya.

Acknowledgments

Specimens were examined from the following museum collections; acronyms used in
the text are listed with each collection. We are indebted to the persons listed with the
collections for allowing us to study specimens in their collections. American Museum
of Natural History (AMNH), New York, Karl F. Koopman; British Museum, Natural
History (BMNH), London; Carnegie Museum of Natural History (CM), Pittsburgh; Field
Museum of Natural History (FMNH), Chicago, Robert Timm and Bruce Patterson;
Museum of Vertebrate Zoology, University of California (MVZ), Berkeley, James L.
Patton; National Museum of Natural History, Smithsonian Institution (USNM), Wash-
ington, D.C., Charles O. Handley, Jr., and Michael Carleton; Papua New Guinea National
Museum and Art Gallery (PNGM), Port Moresby, Simon Poraituk and Paul Banguinan;
Rijksmuseum van Natuurlijke Histoire (RMNH), Leiden, Chris Smeenk; and Zoological
Museum (ZMA), Amsterdam, Peter van Bree.

The field work in Papua New Guinea which resulted in this contribution was part of
a joint Carnegie Museum of Natural History, University of Maryland School of Medicine,
Bishop Museum, Wau Ecology Institute, and Papua New Guinea National Museum and
Art Gallery expedition. Financial support came, in part, from the M. Graham Netting
Research Fund, Carnegie Museum of Natural History, established by a grant from the
Cordelia S. May Charitable Trust, and Grant AI-04242 from the National Institutes of
Health, Washington, D.C., to the Department of Microbiology, University of Maryland
School of Medicine, Baltimore, MD.

Financial and logistical support for the field work was made available by the following
individuals and institutions: Robert Traub, University of Maryland School of Medicine,
Baltimore; Frank J. Radovsky, Alan C. Ziegler, and the late J. Linsley Gressitt (also then

1983

SCH LITTER ET KV.. — ASELLISCUS TAXONOMY

357

Director, Wau Ecology Institute) of the Bernice P. Bishop Museum, Honolulu; Harry
Sakulas and Allen Allison of the Wau Ecology Institute, Wau, Papua New Guinea; and
Simon Poraituk and Paul Banguinan of the Papua New Guinea National Museum and
Art Gallery, Port Moresby. Assistance in the field was supplied by Paul Wanga of the
Papua New Guinea National Museum and Art Gallery and Abid Beg Mirza and Gari
Binyuo of the Wau Ecology Institute. Permission to conduct research in Papua New
Guinea was received from the Institute of Papua New Guinea Studies. Permits were
obtained from Mick Raga, Wildlife Division, Office of Environment and Conservation,
Port Moresby. We are grateful to all of these persons and institutions.

Alan C. Ziegler, Frank S. Radovsky, and Robert Traub reviewed an early draft of the
manuscript; we thank them for doing so.

Literature Cited

Corbet, G. B., and J. E. Hill. 1980. A world list of mammalian species. British Mus.

(Nat. Hist.)/Cornell Univ. Press, London/Ithaca, viii + 226 pp.

Dammerman, K. W. 1929. Fauna Buruana. Mammalia. Treubia Suppl., 7:149-164.
Davis, W. B. 1973. Geographic variation in the fishing bat, Noctilio leporinus. J.
Mamm., 54:862-874.

Dixon, W. J., and M. B. Brown (eds.). 1977. BMDP-77: Biomedical Computer Pro-
grams, P-series. Univ. California Press, Berkeley, xiii + 880 pp.

Dobson, G. E. 1877. On a collection of Chiroptera from Duke-of-York Island and the
adjacent parts of New Ireland and New Britain. Proc. Zool. Soc. London, 1877:
114-123.

. 1878(2. Additional notes on the Chiroptera of Duke-of-York Island and the

adjacent parts of New Ireland and New Britain. Proc. Zool. Soc. London, 1878:
314-318.

. 1 8 78Z). Notes on recent additions to the collection of Chiroptera in the Museum

d’Histoire Naturelle at Paris, with descriptions of new and rare species. Proc. Zool.
Soc. London, 1878:873-880.

. 1878c. Catalogue of the Chiroptera in the collection of the British Museum.

Trustees British Mus., London, xlii + 567 pp.

Hayward, B. J. 1970. The natural history of the cave bat, Myotis velifer. West. New
Mexico Univ. Res. Sci., 1:1-74.

Hill, J. E. 1968. Notes on mammals from the islands of Rennell and Bellona. Pp. 53-
60, in The natural history of Rennell Island, British Solomon Islands (T. Wolff, ed.),
Danish Sci. Press Ltd., Copenhagen, vol. 5.

. 1983. Bats (Mammalia: Chiroptera) from Indo-Australia. Bull. British Mus.,

(Nat. Hist.) Zool., 45:103-208.

Koopman, K. F. 1 982. Results of the Archbold Expeditions No. 1 09. Bats from Eastern
Papua and the East Papua Islands. Amer. Mus. Novit., 2747:1-34.

Laurie, E. M. O., and J. E. Hill. 1954. List of land mammals of New Guinea, Celebes
and adjacent islands 1758-1952. British Mus. (Nat. Hist.), London, 175 pp.

Long, C. A. 1968. An analysis of patterns of variation in some representative Mam-
malia. Part 1. A review of estimates of variability in selected measurements. Trans.
Kansas Acad. Sci., 71:201-227.

Martin, C. O., and D. J. Schmidly. 1982. Taxonomic review of the pallid bat,
Antrozous pallidus (Le Conte). Spec. Publ. Mus., Texas Tech Univ., 18:1-48.
McKean, J. L. 1 972. Notes on some collections of bats (Order Chiroptera) from Papua-
New Guinea and Bougainville Island. Tech. Paper Div. Wildlife Res., C.S.I.R.O.,
Australia, 26:1-35.

Peters, W., and G. Doria. 1880. Enumerazione dei Mammiferi raccolti da O. Beccari,
L. M. D’ Albertis ed A. A. Bruijn nella Nuova Guinea propriamente detta. Ann. Mus.
Civ. Stor. Nat. Genova, 16:665-707.

358

Annals of Carnegie Museum

VOL. 52

Power, D. M. 1970. Geographic variation of red-winged blackbirds in central North
America. Univ. Kansas Publ., Mus. Nat. Hist., 19:1-83.

Sanborn, C. C., and A. J. Nicholson. 1950. Bats from New Caledonia, the Solomon
Islands, and New Hebrides. Fieldiana Zool., 31:313-338.

Smith, J. D. 1972. Systematics of the chiropteran family Mormoopidae. Misc. Publ.,
Univ. Kansas Mus. Nat. Hist., 56:1-132.

Smith, J. D., and C. S. Hood. 1981. Preliminary notes on bats from the Bismarck
Archipelago (Mammalia: Chiroptera). Science in New Guinea, 8:81-121.

SoKAL, R. R., AND F. J. Rohlf. 1 969. Biometry: the principles and practise of statistics
in biological research. W. H. Freeman and Co., San Francisco, xii -I- 776 pp.

SwANEPOEL, P., AND H. H. Genoways. 1979. Morphometries. Pp. 13-106, in Biology
of bats of the New World family Phyllostomatidae, Part III (R. J. Baker, J. K.
Jones, Jr., and D. C. Carter, eds.). Spec. Publ. Mus., Texas Tech Univ., 16:1-441.

^ V,.

i.

J"

*

■t'

J

,n- ,

'iM.

ISSN 0097-4463

7. 73

AN N ALS

0/ CARNEGIE MUSEUM

CARNEGIE MUSEUM OF NATURAL HISTORY

4400 FORBES AVENUE « PITTSBURGH, PENNSYLVANIA 15213
VOLUME 52 16 SEPTEMBER 1983 ARTICLE 16

SOUTH ASIAN MIDDLE EOCENE MOERITHERES
(MAMMALIA: TETHYTHERIA)

Robert M. West*

Director, Carnegie Museum of Natural History

Abstract

Anthracobune and Lammidhania from the middle Eocene of South Asia (Pakistan
and India) are confirmed as belonging to the Moeritheriidae. Newly collected material
permits more complete description of A. pinfoldi, additional comparisons with late
Eocene northern African Moeritherium, and some minor systematic adjustments. It is
now clear that there are two distinct species of Anthracobune. The Moeritheriidae remain
of uncertain position within the Tethytheria.

Introduction

The moeritheres, distinctive subungulate mammals, have long been
considered an exclusively African late Eocene to early Oligocene ra-
diation with possible relationships to the Proboscidea. Discovery of
new middle Eocene specimens from Pakistan and reinterpretation of
previously-described materials from Pakistan and India now confirm
their presence also in the middle Eocene of South Asia (India and
Pakistan). This paper reviews and illustrates the South Asian moeri-
theres, describes the new material, and amplifies the reasons for the
assignment to the Moeritheriidae.

Abbreviations:

AMNH —American Museum of Natural History, New York.

H-GSP —Howard University-Geological Survey of Pakistan project.

BM(NH) — British Museum (Natural History), London.

YPM —Yale Peabody Museum, New Haven, Connecticut.

' Former address: Geology Section, Milwaukee Public Museum, Milwaukee, WI 53233.
Submitted 21 March 1983.

359

360

Annals of Carnegie Museum

VOL. 52

LUVP —Lucknow University Vertebrate Paleontology Collection, Lucknow, India.

Munich — Institut fur Palaontologie und historische Geologic, Munich, Federal Re-
public of Germany.

L —maximum length.

W —maximum width.

Wa —maximum anterior width.

Wp —maximum posterior width.

Previous Studies

African moeritheres were first found along the margin of the Fayum
depression in northern Egypt (Andrews, 1901, 1906). Subsequent dis-
coveries in Libya, Mali, and Senegal have given late Eocene and early
Oligocene Moeritherium (the only African genus) a substantial northern
African distribution (Coppens and Beden, 1978; Tassy, 1981).

Eocene vertebrates have been known from northern South Asia (Fig.
1) since 1877, but detailed work began only in 1940 when Pilgrim
described two of the three taxa here regarded as Asian moeritheres. In
the small collection made by Attock Oil Company geologists from north
of the village of Basal in what is now known as the Punjab Province
of Pakistan, Pilgrim recognized the new genus Anthracobune with two
species {A. pinfoldi and A. daviesi) and a third (“^.” wardi) of possible
affinities. The holotype oiA. pinfoldi is two dentary fragments (BM(NH)
M 15792) (Fig. 2a), and Pilgrim also placed three other lower jaw
fragments in that species. A. daviesi was defined from a single maxillary
fragment (BM(NH) M 15795) (Fig. 2b) containing what Pilgrim re-
garded as P^ and P^. The holotype of “yl.” wardi is the posterior half
of a lower molar (BM(NH) M 15799) (Fig. 2c).

In 1958, Dehm and Oettingen-Spielberg added a further species, Pil-
grimella pilgrimi, based on an isolated upper molar (Munich 1956 II
20) (Fig. 4a) to this group of what were then regarded as anthracotheres.
Sahni and Khare (1973), working in rocks of equivalent age and similar
facies in Jammu and Kashmir, India, recovered a maxilla with P^-
M^ (LUVP 15006) (Fig. 4b) which they assigned to P. pilgrimi.

Gingerich (1 977) restudied Pilgrim’s materials in the British Museum
(Natural History), located several specimens that had not been dis-
cussed by Pilgrim, and determined that the holotype of “T.” wardi
should be placed in a new genus, Lammidhania; he left this genus as
well as Anthracobune and Pilgrimella in the Anthracotheriidae. Gin-
gerich also synonymized A. pinfoldi and A. daviesi on the basis of size
and occurrence, as none of the specimens in Pilgrim’s sample preserved
the same teeth. Thus A. pinfoldi became the first of this group to be
known from both upper and lower dental materials. Gingerich et al.
(1979) also retained all three genera in the Artiodactyla, but removed
BM(NH) 32168 from Lammidhania to cf Anthracokeryx.

Coombs and Coombs (1979) analyzed the ordinal affinities of Pil-

1983

West— Asian Eocene Moeritheres

361

Fig. 1. — Distribution of Eocene sedimentary rocks of western India and northern Pa-
kistan. 1. Chorlakki and Kohat area, Northwest Frontier Province. 2. Basal and Ganda
Kas area, Kala Chitta Hills, Punjab Province. 3. Subathu Formation, Jammu and Kash-
mir State.

grimella based on a detailed study of a cast of LUVP 15006; they then
removed Pilgrimella from the Artiodactyla and concluded that it is a
perissodactyl not assignable to any known family.

West (1980), working with several newly-collected specimens from
north of Basal (the product of field work by the H-GSP project in area
2 of Fig. 1), described a lower dentition (H-GSP 1981) (Figs. 5 and 6)
readily referable to the same species as LUVP 15006. That lower den-
tition obviously belongs to Anthracobune, and West assigned the new
material, plus P. pilgrimi, to A. pinfoldi, which was placed in the Moeri-
theriidae. A. daviesi was not considered, and it and Lammidhania were
left in the Artiodactyla.

Gingerich and Russell (1981) retained both Anthracobune and PU-
grimella as valid genera and put them, as well as Lammidhania, in the
Moeritheriidae. In the same paper, the material used as the holotype
of the presumed sirenian Ishatherium subathuensis, described by Sahni
and Kumar (1980) from the early or middle Eocene of Jammu and

362

Annals of Carnegie Museum

VOL. 52

Kashmir, was regarded as a partial upper molar of Pilgrimella. Gin-
gerieh and Russell then speeulated on possible relationships between
these South Asian species and the eastern Asian Phenacolophidae.

Field work by the Howard University— Geological Survey of Paki-
stan project in March 1982 produced a second maxilla and an asso-
ciated scapula fragment (H-GSP 82-3 IP) (Fig. 3) here placed in An-
thracobune. Study of these and all related specimens conhrms the
synonymy of A. pinfoldi and A. daviesi as suggested by Gingerich et
al. (1979) as well as the synonymy of Anthracobune and PUgrimella
proposed by West (1980). LUVP 15006 and several other specimens
(Table 1) are the same size as Dehm and Oettingen-Spielberg’s holotype
of Pilgrimella pilgrimi; that species, now regarded as valid {contra West
1980), is placed in Anthracobune.

Systematic Paleontology

Order Uncertain or Proboscidea
Family Moeritheriidae Andrews 1 906
Genus Anthracobune Pilgrim 1940

Synonymy.— Pilgrimella Dehm and Oettingen-Spielberg, 1958.

Emended Approximately two-thirds the size of Moeri-

therium; P3 and P4 have distinct trigonids and two talonid cusps;
lower molars increase in size posteriorly, are transversely lophodont
with deep median valleys, have distinct posterior cingular cusps (dou-
bled on M3), and have a paraconid connate with the metaconid; third
and fourth upper premolars have both labial and lingual cusps; upper
molars, increasing in size posteriorly, are lophodont with well-devel-
oped conules anterior to the major cusps and have no obvious ectoloph.
The lower jaws are pronouncedly shallow anteriorly, with a short sym-
physis which extends posteriorly only to the level of Pj.

Included species.— A. pinfoldi Pilgrim, 1940; A. pilgrimi (Dehm and
Oettingen-Spielberg, 1958).

Distribution. — LatQ early and/or early middle Eocene, northern South
Asia.

Anthracobune pinfoldi Pilgrim 1940
Figs. 2a, b; 3a, b

Synonymy.— Anthracobune daviesi Pilgrim 1940.

Holotype. M 15792 plus M 15794 (added by Gingerich
in 1977).

Referred “BM(NH) M 15795 (holotype of A. daviesi), M 15793, 32169,

and H-GSP 82-3 IP.

1983

West— Asian Eocene Moeritheres

363

Table \.— Measurements, in millimeters, of selected specimens of Anthracobune.

A. pinfoldi, BM(NH) M 15795, holotype of “^4. daviesi"

P2

L approx.

15.5

W

18.7

P3

L

19.0

W

23.1

pinfoldi, H-GSP 82--31P

P3

L

19.1

W

23+

p4

L

20.3

W

26+

L

25.0

W

ant

26+

w

'''' post

22.5

M2

L

26.3

30+

w

post

25+

M3

L

27+

Wan,

27.6

w

post

18+

pilgrimi, Munich 1956 II 20, ]

M‘

L

18.2

w

ant

18.4

w

post

19.6

pilgrimi, Munich 1956 II 21

M2

L

21.0

W

ant

19.4

w

post

14.5

pilgrimi, H-GSP 538

M2

L

21.0

W

ant

22.0

w

post

17.5

. pilgrimi, LUVP 15006 (cast)

P2

L

17.6

W

12.2

p3

L

17.3

W

17.7

p4

L

15.7

W

18.6

M‘

L

18.1

Wan,

16.8

^ post

14.0

M2

L

19.0

W

’’ ant

20.0

w

post

17.0

364

Annals of Carnegie Museum

VOL. 52

Fig. 2.—Anthracobune holotypes; all scale units equal 1 cm. a) A. pinfoldi, BM(NH)
M 15792; b) A. daviesi, BM(NH) M 15795; c) ‘M.” wardi, BM(NH) M 15799.

Type locality. — ^onh of Basal, Attock District, Punjab Province,
Pakistan (West and Lukacs, 1979).

Diagnosis. — TdLVgQ species of Anthracobune (P^ length 19 mm;
length 26+ mm; M3 length 17-19.5 mm); upper premolars with single
labial cusps.

Remarks.— Tht recovery of H-GSP 82-3 IP from H-GSP locality
146 (West and Lukacs, 1979) finally allows proper identification of the
teeth preserved in BM(NH) M 15795, the holotype of A. daviesi, as P^
and P^. Both BM(NH) M 15795 and H-GSP 82-3 IP have only a single
labial cusp on P^, in contrast with the incipiently paired cusps in ma-
terial regarded below as A. pilgrimi. Although the sample size now
numbers only five specimens, the size differences between this suite

1983

West— Asian Eocene Moeritheres

365

Fig. 3.—Anthmcobune pinfoldi, H-GSP 82-3 IP; all scale units equal 1 cm. a) Occlusal
view; b) lateral view.

and A. pilgrimi is substantial enough (15% to 30% in various dental
measurements) to assure the reality of these two species of Anthraco-
bune. Because of the difficulties of working with such small samples,
West (1980) inappropriately extended A. pinfoldi to include physically
small specimens which here are excluded.

The dentition of H-GSP 82-3 IP, the first known upper cheek teeth
of A. pinfoldi, is similar to that of A. pilgrimi which has previously been
described (Sahni and Khare, 1973; Coombs and Coombs, 1979). The
third premolar is three-rooted and high-crowned. There is a single high
labial cusp and a lower lingual cusp which are joined by a low anterior

366

Annals of Carnegie Museum

VOL. 52

crest. The degree of enamel spalling allows recognition of a low cin-
gulum only posteriorly.

also is three rooted and, although preservation is poor, probably
was not as high crowned as P^. The single external cusp gives olf low
anterior and posterior crests to the vicinity of the external ends of the
flat anterior and posterior cingula. There apparently is no external
cingulum.

The first upper molar is four rooted and noticeably wider anteriorly
than posteriorly. The bulbous paracone contributes to this. The external
cingulum, discontinuous along the posterior part of the paracone, ex-
tends anteriorly into a small parastyle. The “ectoloph” between the
paracone and metacone is very low, at essentially the same level as a
medial elevation that extends from the posterior corner of the proto-
cone to the anterior side of the metaconule. The metaconule is ante-
riorly situated. The internal and posterior cingula are discontinuous
around the hypocone, unlike the condition in A. pilgrimi.

is much less worn than but is structurally very similar. Its
cingulum, prominent except around the posterior part of the paracone,
extends anteriorly into a small parastyle. Narrow crests from the pos-
terior flank of the paracone and the anterior flank of the metacone
converge on a tiny mesostyle. As with the posterior part of the
tooth is substantially narrower than the anterior part. Both conules are
large; the metaconule is situated well anterior of a metacone-hypocone
line and is strongly connected to each. The paraconule is forward of a
paracone-protocone line and is similarly well joined to both conules.
The major cusps are conical.

The upper third molar shows the same degree of wear as M^. The
anterior cusps, paracone, paraconule and protocone, are all approxi-
mately equal in size. There is a strong anterior cingulum and the ex-
ternal cingulum is slightly expanded in the position of a parastyle. The
valley between the paracone and metacone is deeper than that between
the protocone and metaconule. Most of the hypocone is broken away,
but it seems to have been a low cingular cusp immediately posterior
to the metacone. The internal cingulum, discontinuous around the
protocone, also is broken away in the hypocone area.

A fragment of the proximal end of the right scapula was found in
direct association with the maxilla. The glenoid cavity is oval and
shallow with rounded margins. Anteriorly it extends into a bulbous
notched coracoid process. The coracoid edge is concave, approximately
parallel to the spine. The postero-internal corner of the glenoid cavity
is broken away, but it appears from the small unbroken area that the
coracoid border also is concave. This indicates that the distal part of
the scapula, the blade, is broadly flaring. The spine rises abruptly about
3.5 cm from the margin of the glenoid cavity. Although it is broken.

1983

West— Asian Eocene Moeritheres

367

B

Fig. A.—PUgrimella pilgrimi (=Anthracobune pilgrimi)-, all scale units equal 1 cm. a)
Munich 1956 II 20, holotype; b) LUVP 15006.

the spine is slanted slightly toward the glenoid border, producing a
deeper and sharper fossa than on the coracoid side.

Anthracobume pilgrimi (Dehm and Oettingen-Spielberg 1958)
Figs. 4a, b; 5; 6

Synonymy.— Pilgrimella pilgrimi Dehm and Oettingen-Spielberg 1958.
Pilgrimella pilgrimi Sahni and Khare 1973.

Pilgrimella pilgrimi Coombs and Coombs 1979.
Anthracobune pinfoldi (part) West 1980.

368

Annals of Carnegie Museum

VOL. 52

Fig. 5.—Anthmcobune pilgrimi, H-GSP 1981, occlusal view; scale units equal 1 cm.

Holotype. — Munich 1956 II 20.

Referred specimens. -LUVF 15006; H-GSP 538, 568, 1981, 1975; Munich 1956 II
21.

Type locality. — r>chm loc. 24 (==H-GSP loc. 67 of West and Lukacs
1979), north of Basal, Attock District, Punjab Province, Pakistan.

Diagnosis. — Small species of Ant hracobune (P^ length 16-17 mm;
length 16-21 mm; M3 length 25-26 mm); upper premolars with
two labial cusps.

Remarks. — This is the better known of the two species of Ant hra-
cobune. The entire postcanine lower dentition is included in a pair of
well-preserved dentaries (H-GSP 1981), and the upper teeth (P^-M^)
are preserved in LUVP 15006. That maxilla was described by Sahni
and Khare (1973) and reviewed by Coombs and Coombs (1979). The
dentition of H-GSP 1981 was fully described previously (West, 1980),
but when that paper was written the preparation of the dentary bones
was incomplete. Subsequently, they have been fully cleaned and freed
of matrix and are illustrated in Figs. 5 and 6. Some of the features

1983

West— Asian Eocene Moeritheres

369

Fig. 6~-Anthracobune pilgrim, H-GSP 1981, lateral view; scale units equal 1 cm.

described earlier, including the shallow symphyseal region and alveoli
for the anterior teeth are now much clearer. In addition, a prominent
bulge extends along the lingual side of the dentary bones from the
posterior premolars anteriorly almost to the symphysis on both sides.
The bone in the symphysis is approximately as thick as that of the
bulge. The lingual surface of the symphysis parallels the flattened labial
surface, emphasizing the rapid anterior shallowing of the jaw. Detailed
preparation of the incisor region now indicates the probable presence
of three procumbent incisors, all smaller than the canine.

Lammidhania GmgQTich 1977

Synonymy. — '"Anthracobune” Pilgrim 1940.

Diagnosis. — Size (M3 length 24 mm) and general morphology similar
to Anthracobune; lower molars relatively narrower than in Anthraco-
bune; three lower premolars.

Included species— L. wardi (Pilgrim 1940).

Lammidhania wardi (Pilgrim 1940)

Fig. 2c

Synonymy.— ''‘Anthracobune’'’ Pilgrim 1940.

i7o/o/j;/?G-BM(NH) M 15799.

Referred specimens. — M.-GS'F 1000, 982, 1633.

370

Annals of Carnegie Museum

VOL. 52

Diagnosis.— As for the genus.

Remarks.— The dilferences between Lammidhania wardi and An-
thracobune pilgrimi are primarily in the area of premolar number and
morphology and relative robustness of the cheek teeth. These characters
are minimally adequate for generic distinction {Lammidhania is more
derived than Anthracobune in the loss of a premolar, probably Pj) but
the close higher-level relationship between these genera is clear. West
(1980) erred when he suggested that the premolar condition of Lam-
midhania was indicative of an artiodactyl relationship and the molar
structure was a convergence with Anthracobune. My current preference
is to regard Lammidhania as a moerithere, closely related to and doubt-
fully separate from Anthracobune.

Moerithere Affinities of Anthracobune and Lammidhania

The initial placement of Anthracobune in the Moeritheriidae (West,
1980) relied to a significant extent on mandibular and dental characters
readily seen in H-GSP 1981 and LUVP 15006. These included rela-
tively wide lower cheek teeth, prominent lophodonty and hypoconu-
lids, a procumbent symphyseal area, complex premolars, and an abruptly
rising ascending ramus. The absence of an ectoloph in the upper molars,
enlarged conules on both the paraloph and metaloph, and strong cingula
ally the upper teeth with moeritheres. The new data available from
H-GSP 82-3 IP further support moerithere affinities— the nature of the
anterior root of the zygoma with its opening for the anterior orbital
foramen and the distinct enlargement dorsal to M--M^ and the ffaring
scapula with a prominent coracoid process.

Anthracobune is more primitive than Moeritherium in the forward
placement of the symphysis, retention of all the incisors, only mod-
erately enlarged canines, posterior position (adjacent to posterior end
of M3) of the ascending ramus, more molariform P4, smaller M3 heel,
and interruption in the lophs on the lower molars. In the maxilla,
primitive features of Anthracobune are the lack or incomplete sepa-
ration of the paracone and metacone on P^ and P^, the degree of in-
dependence of the molar conules, the relatively great anterior breadth
of the molars, and the posterior position of the downturned and en-
larged maxillary root of the zygomatic arch.

In all these characters, Anthracobune is less derived than is Moer-
itherium, and more similar to the presumed ancestral tethythere stock.
It has fewer similarities with the most primitive known sirenians, as
it possesses a short and shallow mandibular symphysis, only four pre-
molars, and small mental foramina, and lacks any indication of pach-
yostosis. Similarly, the South Asian creatures are more primitive than
the northern African late Eocene-early Oligocene barytheres which

1983

West ”= Asian Eocene Moeritheres

371

have lost the canines, one incisor and Pi, have a very deep and massive
symphysis, and a prominent anteriorly-situated ascending ramus.

Environments

Both the middle Eocene Asian and late Eocene-early Oligocene Af-
rican moeritheres have been collected from rocks deposited in similar
coastal environments. The Anthracobune and Lammidhania materials
for which adequate data are available (those collected near Basal by
the Munich and H-GSP projects) have come from throughout the
Kuldana Formation. That rock unit is composed primarily of fresh-
water calcareous mudstone, fine-grained sandstone, and granular cal-
clithite. Foraminifera-bearing limestone lenses are present, indicative
of the proximity of the fluctuating Tethys shoreline (West and Luckacs,
1979). To the west, in the Chorlakki area of the Kohat Basin (No. 1
on Fig. 1), Eocene rocks produce a similar fauna; Gingerich et al. (1983)
and Wells (1983) regarded these as indicative of coastal environments
surrounding a saline epicontinental sea remnant of the contracting
Tethys.

The faunas accompanying Anthracobune and Lammidhania (West,
1980; Gingerich and Russell, 1981 Gingerich et aL, 1983; and refer-
ences cited therein) include numerous primitive whales as well as typ-
ical terrestrial organisms. Thus both lithological and paleontological
data are compatible with an amphibious habit for the middle Eocene
Asian moeritheres.

The moerithere-bearing beds in the Fayum of Egypt indicate a near-
shore environment (Bown et al, 1982). The late Eocene fossiliferous
unit, the Qasr el-Sagha Formation, ranges from “interdeltaic shallow
marine and littoral” to “deltaic . . . with distributory channel sands”
(Simons, 1968). Overlying it, the Jebel el Qatrani Formation is more
terrestrial, representatWe of a channel floodplain complex. The Qasr
el-Sagha Formation contains numerous moeritheres in association with
archaeocete whales (Andrews, 1906), a situation parallel to that of the
Kuldana Formation in Pakistan.

CO'NCLUSIONS

Three species of primitive moeritheres are now recognized in late
early to early middle Eocene rocks in South Asia. They are represen-
tative of a slightly earlier stage in the evolution of this group than is
African Moeritherium, but existed under ecologic conditions quite sim-
ilar to those of the North African localities.

While these new interpretations expand the moeritheres both geo-
logically and geographically, they still are distinct from the other teth-
ytheres (fide McKenna, 1975). Middle Eocene moeritheres do not, at

372

Annals of Carnegie Museum

VOL. 52

this time, contribute toward more precisely relating moeritheres to the
remaining tethytheres; at the very least, the moerithere-sirenian di-
chotomy occurred earlier in the Eocene (Domning et al., 1982; Sereno,
1982), and the moeritheres may be best regarded as a close sister group
of the Sirenia and of the Proboscidea if they are not properly placed
in that order.

Acknowledgments

Drs. Richard Tedford (American Museum of Natural History), Leo Hickey and John
Ostrom (Yale Peabody Museum), Ashok Sahni (Centre of Advanced Studies in Geology,
Panjab University, Chandigarh, India), and Richard Dehm (Staatsamlung fur Palaon-
tologie und historisch Geologic, Munich) and Mr. Jeremy Hooker (British Museum (Nat-
ural History)) allowed me access to collections in their respective charges. Mr. Hooker
and Drs. Tedford, Philip Gingerich and Margery Coombs provided me with casts. Dr.
Coombs and Dr. Leonard Krishtalka reviewed the manuscript. The illustrations were
prepared by Ms. Susan Speerbrecher and the fossils were cleaned by Mr. Rolf Johnson,
both of the Milwaukee Public Museum.

This contribution is part of the ongoing work of the Howard University —Geological
Survey of Pakistan “Fossil Mammals of Pakistan” project directed by Drs. S. Taseer
Hussain (Howard University) and S. M. Ibrahim Shah (Geological Society of Pakistan).
The field work benefitted from the able services of Mr. Muhammed Arif, Assistant
Director, Geological Survey of Pakistan.

Literature Cited

Andrews, C. W. 1901. Preliminary notes on some recently discovered extinct verte-
brates from Egypt. Part 1. Geol. Mag., IV, 8:400-409.

. 1906. A descriptive catalogue of the Tertiary vertebrates of the Fayum, Egypt.

Brit. Mus. (Nat. Hist.), London, 324 pp.

Bown, T. M., M. J. Kraus, S. L. Wing, J. G. Fleagle, B. H. Tiffney, E. L. Simons,
and C. F. Vondra. 1982. The Fayum primate forest revisited. J. Human Evol.,
11:603-632.

Coombs, M. C., and W. Coombs. 1979. Pilgrimella, a primitive Asiatic perissodactyl.
Zool. J. Linn. Soc., 65:185-192.

CoppENS, Y., and M. Beden. 1978. Moeritheriidae. Pp. 333-335, in Evolution of
African mammals (V. J. Maglio and H. B. S. Cooke, eds.). Harvard Univ. Press,
Cambridge, 641 pp.

Dehm, R., and T. zu Oettingen-Spielberg. 1958. Palaontologische und geologische
Untersuchungen im Tertiar von Pakistan. 2. Die mitteleocanen Saugetiere von Gan-
da Kas bei Basal in Nordwest Pakistan. Abh. Bayer. Akad. Wiss., Math. -Nat. KL,
91:1-54.

Domning, D. P., G. S. Morgan, and C. E. Ray. 1982. North American Eocene sea
cows (Mammalia: Sirenia). Smithsonian Contrib. Paleobiology 52:1-69.
Gingerich, P. D. 1977. A small collection of fossil vertebrates from the middle Eocene
Kuldana and Kohat formations of Punjab (Pakistan). Contrib. Mus. Paleontol.,
Univ. Michigan, 24:190-203.

Gingerich, P. D., and D. E. Russell. 1981. Pakicetus inachus, a new archaeocete
(Mammalia, Cetacea) from the early-middle Eocene Kuldana Formation of Kohat
(Pakistan). Contrib. Mus. Paleontol., Univ. Michigan, 25:235-246.

Gingerich, P. D., D. E. Russell, D. Sigogneau-Russell, J. -L. Hartenberger, S. M.
Ibrahim Shah, M. Hassan, K. D. Rose, and R. H. Ardrey. 1 979. Reconnaissance

1983

West— Asian Eocene Moeritheres

373

survey and vertebrate paleontology of some Paleocene and Eocene formations in
Pakistan. Contrib. Mus. PaleontoL, Univ. Michigan, 25:105-116.

Gingerich, P. D., N. a. Wells, D. E. Russell, and S. M. Ibrahim Shah. 1983. Origin
of whales in epicontinental seas: new evidence from the early Eocene of Pakistan.
Science, 220:403-406.

McKenna, M. C. 1975. Toward a phylogenetic classification of the Mammalia. Pp.
21-46, in Phylogeny of the primates: a multidisciplinar>' approach (W. P. Luckett
and F. S. Szalay, eds.). Plenum, New York, 483 pp.

Pilgrim, G. E. 1940. Middle Eocene mammals from northwest Pakistan. Proc. Zool.
Soc. London, ser. B, 110:127-152.

Sahni, a., and S. K. Khare. 1973. Additional Eocene mammals from the Subathu
Formation of Jammu and Kashmire. J. Paleo. Soc. India, 17:31-49.

Sahni, A., and K. Kumar. 1 980. Lower Eocene Sirenia, Ishatherium subathuensis gen.
et sp. nov. from the type area, Subathu Formation, Subathu, Simla Himalayas, U.P.
J. Paleo. Soc. India, 23-24:132-135.

Sereno, P. C. 1982. An early Eocene sirenian from Patagonia (Mammalia: Sirenia).
Amer. Mus. Novit., 2729:1-10.

Simons, E. L. 1968. Early Cenozoic mammalian faunas, Fayum Province, Egypt. Part
1. African Oligocene mammals: Introduction, history of study, and faunal succes-
sions. Peabody Mus. Bull., 28:1-21.

Tassy, P. 1981. Le crane du Moeritherium (Proboscidea, Mammalia) de I’Eocene de
Dor el Talha (Libye) et le probleme de la classification phylogenetique du genre dans
les Tethytheria McKenna, 1975. Bull. Mus. Nat. Hist. Nature., ser. 4, 3:87-147.

Wells, N. A. 1983. Transient streams in sand-poor redbeds; early-middle Eocene
Kuldana formation of northern Pakistan. Spec. Publ. Internat. Assoc. Sedimentol-
ogists, 6:393-403.

West, R. M. 1980. Middle Eocene large mammal assemblage with Tethyan affinities,
Ganda Kas region, Pakistan. J. Paleo., 54:508-533.

West, R. M., and J. R. Luckas. 1979. Geology and vertebrate-fossil localities. Tertiary
continental rocks, Kala Chitta Hills, Attock District, Pakistan. Milwaukee Public
Mus., Contrib. Biol. Geol., 26:1-20.

Back issues of many Annals of Carnegie Museum articles are
available, and a few early complete volumes and parts are listed
at half price. Orders and inquiries should be addressed to: Pub-
lications Secretary, Carnegie Museum, 4400 Forbes Avenue,
Pittsburgh, Pa. 15213.

73 I

{ ISSN 0097-4463 ,

ANNALS

0/ CARNEGIE MUSEUM

CARNEGIE MUSEUM OF NATURAL HISTORY

4400 FORBES AVENUE » PITTSBURGH, PENNSYLVANIA 15213
VOLUME 52 7 DECEMBER 1983 ARTICLE 17

REVISION OF THE WIND RIVER FAUNAS,

EARLY EOCENE OF CENTRAL WYOMING.

PART 4. THE TILLODONTIA

Richard K. Stucky

Post-doctoral Fellow, Section of Vertel

Leonard Krishtal|a 0 Q 1 9 1983

Associate Curator, Section of Venerate Fossils

Abstract

Two species of Esthonyx—E. bisulcatus and E. acutidens—are recorded from the Lysite
Member, and the Red Creek Facies and Lost Cabin Member, respectively, of the Wind
River Formation. Trogosus, hitherto unknown from this formation, has been recovered
from two localities in the upper part of the Lost Cabin Member. The first occurrence of
Trogosus, along with that of other typical Bridgerian mammals, marks the Wasatchian-
Bridgerian boundary and, in part, defines the Gardnerbuttean (earliest Bridgerian) subage.
A reconstruction of the phylogenetic relationships of North American and Asian tillo-
donts suggests that the North American genera Esthonyx, Megalesthonyx, and Trogosus
may have had independent origins from among Asian tillodonts.

Introduction

Tillodonts were among the first fossil mammals reported from the
Wind River Basin. Of the two species Cope (1880, 1881) named from
the Wind River ¥orm2it\on— Esthonyx acutidens and E. spatularius—
the type of the latter appears to have come from the Bighorn Basin
(Gazin, 1953). Subsequently, excellent material of E. acutidens from
the Boysen Reservoir area. Wind River Basin (White, 1952), was stud-
ied by Gazin (1953) in his major review of the Tillodontia. He formally
recognized E. bisulcatus Cope, 1874 and E. acutidens from the “Lysite”
and “Lost Cabin” beds, respectively, of the Wind River Formation —

Submitted 13 May 1983.

375

376

Annals of Carnegie Museum

VOL. 52

a conclusion corroborated in subsequent reviews (Kelley and Wood,
1954; Guthrie, 1967, 1971)— and suggested that E. acutidens may also
occur in the Lysite Member. Gingerich and Gunnell (1979), in the last
major treatment of tillodont systematics and evolution, did not specify
the biostratigraphic occurrences of either species in particular basins.
They reported E. bisulcatus from Lysitean and Lostcabinian aged ho-
rizons, and E. acutidens from “Lost Cabin beds and equivalents” (p.
146) and early Bridgerian horizons in western North America. Im-
portantly, Trogosus has never been reported from the Wind River
Formation.

Since Guthrie’s (1967, 1971) work, 20 specimens from the Lysite
Member (nine localities), 34 from the Lost Cabin Member (12 local-
ities), and two from the Red Creek Facies (two localities; Stucky, 1983)
have been added to the sample of tillodonts from the Wind River
Formation. Two of the specimens— from the Lost Cabin Member—
represent Trogosus, and the remainder, Esthonyx. Four of the localities
(CM loc. 34; UCM loes. 80101, 81028, 79040), all in the upper part
of the Lost Cabin Member, are earliest Bridgerian (=Gardnerbuttean,
Robinson, 1966) rather than latest Wasatehian, according to biostrati-
graphic evidence presented here and elsewhere (Stucky, 1982, 1983;
Krishtalka and Stucky, 1983). This conclusion is reflected in the “known
occurrence” data in the systematic section below.

Abbreviations in text and tables are as follows: AMNH, American Museum of Natural
History; CM, Carnegie Museum of Natural History; IVPP, Institute of Vertebrate Pa-
leontology and Paleoanthropology, People’s Republic of China; UCM, University of
Colorado Museum; Loc., locality.

Systematics

Order Tillodontia Marsh, 1875
Family Esthonychidae Cope, 1883

Gazin (1953) and Gingerich and Gunnell (1979) have reviewed the
taxonomic history of tillodonts. Five North American genera are gen-
erally vQcognizQd— Esthonyx Cope, 1874; Trogosus Leidy, 1871; An-
chippodus Leidy, 1868; Tillodon Gazin, 1953; Megalesthonyx Rose,
1972. Esthonyx has been reported from the Clarkforkian, Wasatehian,
and earliest Bridgerian, Trogosus and Tillodon only from the Bridg-
erian, and Megalesthonyx from the late Wasatehian. The age of An-
chippodus, although suggested to be Bridgerian (Gazin, 1953), cannot
be determined from present evidence.

One esthonychid, E. munieri (Lemoine, 1889) is known from the
Sparnacian or Cuisian of the Paris Basin, and four have been reported
from the early Tertiary of China. Two of these, Lofochaius Chow et
al., 1973, and Meiostylodon Wang, 1975, are early or middle Paleocene

1983

Stucky and Krishtalka—Wind River Faunas, 4

377

in age, (Li and Ting, 1983), whereas Kuanchuanius Chow, 1963, and
Adapidium Young, 1937, come from middle and late Eocene horizons,
respectively. Basalina Dehm and Oettingen-Spielberg, 1958, from the
middle Eocene of Pakistan, may be an esthonychid (Gingerich and
Gunnell, 1979; Lucas and Schoch, 1981).

Est hony X CopQ, 1874

Gazin (1953) recognized five North American species of Esthonyx,
which from stratigraphically oldest to youngest are: E. granger! Simp-
son, 1937; E. latidens Simpson, 1937; E. spatularius Cope, 1880 (all
Ciarkforkian and Graybullian, Bighorn Basin); E. bisukatus Cope, 1874
(Graybullian and Lysitean, Bighorn., Wind River, San Juan and Pice-
ance basins); E. acutidens Cope, 1881 (?Lysitean and Lostcabinian,
Bighorn, Wind River, Washakie and Piceance basins). He also iden-
tified Esthonyx sp. from the Lostcabinian part of the Huerfano For-
mation, the Cathedral Bluffs Tongue of the Wasatch Formation and the
Lysitean of the Knight (^Wasatch) Formation.

Gingerich and Gunnell (1979) added two new species of Esthonyx—
E. xenicus and E. ancylion— from the Ciarkforkian of the Bighorn
Basin, synonymized E. latidens with E. grangeri, and redefined the
hypodigms of the other North American species. Part of the hypodigm
of E. xenicus includes specimens previously referred to E. spatularius
and E. bisukatus. They posited two lineages of Esthonyx in Ciarkfor-
kian and Wasatchian strata of western North America. Species in the
first lineage— xenicus-E. ancylion-E. grangeri— l^ck symphysial fu-
sion, retain a double-rooted P2, have non-overlapping stratigraphic
ranges, and increase in size anagenetically. Species composing the sec-
ond lineage— spatularius-E. bisukatus-E. acutidens— hmc a fused
mandibular symphysis, a single-rooted P2, and show an anagenetic
increase in size and hypsodonty. According to Gingerich and Gunnell
(1979)— spatularius and E. bisukatus do not overlap in range; spec-
imens that are intermediate in age are also morphologically interme-
diate and tentatively referred to the former, with the provision that
they may represent a new, intermediate species; and finally, the rela-
tionship between E. bisukatus and E. acutidens is less clear, because
they are contemporaneous in the middle and upper part of the Hep-
todon Range-Zone of the Bighorn Basin (Schankler, 1980).

The E. xenicus lineage begins in the early Ciarkforkian, and the E.
spatularius one at the onset of the Wasatchian. When the two lineages
are compared, there is significant overlap in size between E. spatularius
and E. xenicus-E. ancylion, and between E. bisukatus and E. ancylion.
Thus, it is difficult to identify specimens of similar size, unless they
preserve the mandibular symphysis and/or anterior premolars. Ac-
cordingly, we suggest additional diagnostic features, given that the fig-

378

Annals of Carnegie Museum

VOL. 52

ures and descriptions in Gingerich and Gunnell (1 979) accurately reflect
the range of intraspecific morphologic variation. The E. xenicus-an-
cylion-grangeri group may be characterized by inflated metaconids,
appressed paraconids and metaconids and low paracristids on Mi_3,
and very narrow stylar shelves and salients on In contrast, the

E. spatularius-bisulcatus-acutidens group has well-separated paraco-
nids and metaconids, an unexpanded metaconid, and a high paracristid
on Mi_3, and an expanded stylar shelf and salients on These

characters may aid in identifying specimens that lack the diagnostic
anterior portions of the lower dentition.

The E. xenicus-ancylion-grangeri lineage, known only from the
northern part of the Bighorn Basin (including the Clark’s Fork Basin),
appears to be a chronospecies that increases in size. The divisions
between the species are stratigraphic and necessarily arbitrary.

On the other hand, the E. spatularius-bisulcatus-acutidens clade is
not necessarily one chronospecies. The first two species compose a
morphocline of increasing size and hypsodonty through time, and in-
dividual specimens are indistinguishable except stratigraphically. In
agreement with Bown (1979), they are here considered conspecific. E.
bisulcatus-E. acutidens show similar anagenetic change, but their ap-
parent co-occurrence at a number of horizons in the Bighorn Basin
(Schankler, 1980) as well as non-overlapping ranges in size in the Wind
River Basin (see below) imply that, for the present, they should be
retained as separate species. Continued analyses of the Bighorn Basin
material and denser samples frorp the Wind River Basin should clarify
this issue.

Gingerich and Gunnell (1979) have attempted to define statistically
the Clarkforkian and Wasatchian species of Esthonyx. They (1 979: 1 28)
derive a 0.095 standard deviation value (for Ln [L x W] of Mj) from
the “largest sample” of Esthonyx and use it to compute a diagnostic
difference of 3.3 standard deviation units between the type specimens
of E. xenicus and E. ancylion, and 5. 1 standard deviation units between
the types of the latter and E. grangeri. The three type specimens are
compared to one another rather than to the descriptive statistics of any
sample. In this light, these comparisons, although interesting, appear
to violate principles of statistical analysis. The standard deviation of
a sample is not applicable as a model to any other sample, population,
or individual. Also, there are no statistical procedures that test for the
significance of the difference between individual specimens, measure-
ments of which can only be compared to the mean of a population or
statistical sample (by means of a Z-test) (Simpson et al., 1960: 173;
Sokal and Rohlf, 1969).

The assignment in Gingerich and Gunnell (1979) of a Lostcabinian
age to the San Jose tillodont, E. bisulcatus, has been revised. Recent

1983

Stucky and Krishtalka— Wind River Faunas, 4

379

analyses of the AMNH (Lucas et al, 1981) and CM (unpublished data)
collections from the Regina and Tapicitos members of the San Jose
Formation found evidence that the San Jose faunas are late Graybullian
to Lysitean— the absence of the index fossil Lambdotherium, as well
as Shoshonius, and the presence of a host of typical early to middle
Wasatchian mammals (Ambloctonus, Didymictis protenus, Ectoganus,
Apheliscus, Pelycodus jarrovii, Navajovius, Chriacus, Anacodon, Horn-
ogalax).

Esthomyx bisulcatus Cope, 1874
(Fig. 1; Tables 1, 3)

Referred specimens.— Ymiml dentary with P4-M2— CM 39205; P3, dP4— -CM 37053;
M,^CM 39941, 39645, 39902, 37052; M2-CM 39160, 37054, 22660 (2 teeth);

CM 36117, 37056, 28722, 53785; lower molar fragments—CM 37055, 35929, 19902;
partial maxilla with M^-^-^CM 39644; M^-^-CM 54155; upper molar fragments-^-CM
35978, 35979, 35980.

Localities. -CM Iocs. 797, 799, 802, 806, 807, 927, 929, 931, 1091
(Lysitean, Lysite Member).

Known distribution.— emly Wasatchian— Hoback and Powder River
basins (Wasatch Fm.), Wyoming; early to late Wasatchian— Bighorn
Basin (Willwood Fm.X Wyoming; middle Wasatchian— Wind River
Basin (Wind River Fm.), Wyoming, San Juan Basin (San Jose Fm.),
New Mexico.

Description and discussion.— The known morphology of E. bisul-
catus is described in Gazin (1953) and Gingerich and Gunnell (1979).
In the Wind River Basin this species is restricted to, but occurs through^

Table \.~ Dimensions of lower teeth o/Esthonyx bisulcatus from the Lysite Member,
Wind River Formation, Wyoming.

specimen

no.

P3

! .oral-

P4

M,

M2

M3

ity L W

L W

L

w

L

w

L W

CM 37053

807 6.1 3.8

8.7 5.3

CM 39205

931

7.8 5.7

7.8

6.5

8.6

6.8

CM 39941

797

7.5

6.1

CM 39645

802

7.8

6.2

CM 39902

797

7.8

6.2

CM 37052

927

8.0

6.0

CM 39160

806

8.5

6.5

CM 22660

799 -

8.5

6.5

CM 22660

799

8.5

6.5

CM 53785

1091

8.4

6.1

CM 37054

807

6.4

CM 36117

929

10.0 5.2

CM 37056

929

10.0

CM 28722

807

5.4

380

Annals of Carnegie Museum

VOL. 52

Fig. \.-Esthonyx bisulcatus. A) CM 39644, LM‘-^; B) CM 39205, LP4-M2, and CM
36117, LM3; all to same scale (1 cm).

out, the Lysite Member of the Wind River Formation. The absence of
E. bisulcatus from localities in the Red Creek Facies (Stucky, 1983;
Stucky and Krishtalka, 1982) of the Wind River Formation in the Red
Creek-Deadman Butte area is coincident with the Lostcabinian age
assigned to these sediments.

In E. bisulcatus, compared to Wind River E. acutidens, the teeth
are significantly smaller (no overlap in range), P4-M3 are less hypsodont
and bear a less robust metaconid, P4 has a metastylid, Mi_3 have a
stronger metastylid, and P'^-M^ have narrower stylar shelves with less
expanded parastylar and metastylar salients.

1983

Stucky and Krishtalka—Wind River Faunas, 4

381

Fig. 2.—Esthonyx acutidens. A) CM 22311, RDP'*, and CM 42127, LP'‘-M*; B) CM
30943, LM'-\ and CM 22310, LM^; C) CM 35869, LP4, and CM 22308, LM2_3; all to
same scale ( 1 cm).

382

Annals of Carnegie Museum

VOL. 52

Table 2.— Dimensions of lower teeth o/Esthonyx acutidens from the Lost Cabin Member,
Wind River Formation, Wyoming.

Specimen

no.

Locality

P4

M,

M

2

M3

L

w

L

w

L

w

L

w

CM 35869

34

8.8

6.1

CM 35870

34

9.0

6.2

CM 42131

34

8.1

5.2

8.8

6.6

CM 22309

34

8.9

7.0

CM 37236

34

8.6

7.0

CM 30941

34

8.8

6.6

CM 30942

34

8.7

6.5

UCM 45384

80089

8.7

6.6

9.3

6.0

CM 22312

90

10.3

7.2

CM 22312

90

9.1

7.1

CM 22312

90

9.8

7.5

CM 30940

34

9.4

6.8

CM 37235

34

10.0

7.0

CM 36457

91

10.0

7.9

CM 37238

34

9.5

8.0

CM 42130

1077

10.9

7.5

CM 22308

34

9.0

7.8

10.8

5.9

CM 55193

34

11.5

5.7

CM 37239

34

7.6

None of the specimens referred here preserves the mandibular sym-
physis or the anterior dentition— features used by Gingerich and Gun-
nell (1979) to distinguish between the two lineages of tillodonts. How-
ever, the teeth exhibit the stylar shelf development and lower molar
hypsodonty that is typical of E. bisulcatus and E. acutidens. They lack
the anteriorly expanded metaconid that appears to be characteristic of
the E. xenicus-ancylion-grangeri chronospecies.

Esthonyx acutidens Cope, 1881
(Fig. 2; Tables 2-3)

Referred specimens. dentary with Mi_2— CM 21232; M2_3“CM 22308; P4~
M,-UCM 45384; P4, M,-CM 42131; P4-CM 35869, 35870; M,-CM 22309, 37236,
30941, 30942; M2-CM 22312, 30940, 37235, 37238, 36457, 42130; M3-CM 37239,
55193; lower molar fragments— CM 21131, UCM 46397; partial maxilla with P^-M‘ —
CM 42127; M'-^-CM 30943; M^-^-CM 36477; ?associated dP^ M'-CM 22311; I/,
M/-UCM 45528; P-'-CM 30939, 55192; M‘-CM 22312, 42128, 42133; M^-CM
22312, 37237, 21113, 35841, 55191; M^-CM 22310 (2 teeth), 30938, 22313, 55190,
55491; fragmentary upper molars— CM 42129, UCM 46368, 44321 (M/, /M).

Localities. — CM. Iocs. 91, 98, 1048, UCM loc. 80063 (Lostcabinian,
Lost Cabin Member); UCM Iocs. 80089, 81009 (Lostcabinian, Red
Creek Facies); CM loc. 34, UCM loc. 80101 (Gardnerbuttean, Lost

1983

Stucky and Krishtalka—Wind River Faunas, 4

383

Table 'i.— Dimensions of upper teeth o/Esthonyx and Trogosus from the Wind River

Formation, Wyoming.

Specimen

no.

P M'

M2

Locality

L

W L W

L

w

L

W

Esthonyx acutidens

CM 55192

34

9.0

n.i

CM 30939

34

9.0

11.3

CM 42127

34

8.8

12.0

CM 22311

34

8.4

8.6

CM 42128

34

8.2 12.6

CM 30943

34

8.5 12.3

8.9

13.4

CM 22312

90

9.5 13.9

9.7

15.7

CM 35841

34

9.7

14.2

CM 55191

34

8.8

14.2

CM 22313

91

8.0

15.6

CM 36477

34

7.6

12.9

CM 30938

34

7.5

13.4

CM 22310

34

7.3

13.5

CM 42131

34

Esthonyx bisulcatus

7.2

13.0

CM 39644

802

8.1

13.0

7.0

12.4

CM 54155

1091

Trogosus sp.

7.2

12.7

UCM 46536

81028

22.7

36.1

Cabin Member); CM 90, 857, 1077 (Lostcabinian or Gardnerbuttean,
Lost Cabin Member).

Known distribution. ~h3.tQ Wasatchian— Bighorn Basin (Willwood
Fm.), Green River Basin (Wasatch Fm.), Wyoming; late Wasatchian
to early Bridgerian— Wind River Basin (Wind River Fm.), Wyoming,
Huerfano Basin (Huerfano Fm.), Colorado.

Description and Discussion. — \n the Wind River Basin this species
is recovered only from the Lost Cabin Member and Red Creek Facies
of the Wind River Formation, which, in combination with the restric-
tion of E. bisulcatus to the Lysite Member, may be important bio-
stratigraphically.

As was the case with E. bisulcatus, none of the material of E. acu-
tidens referred here preserves the mandibular symphysis or teeth an-
terior to P4. Referral of the material to E. acutidens is based on di-
agnostic criteria on PV4-MV1 outlined earlier. Compared to E.
bisulcatus, the teeth of E. acutidens from the Wind River Formation
are significantly larger, P4 lacks a metastylid, Mi_3 have weaker meta-
stylids, and P4-M3 are more hypsodont with lingually robust meta-
conids; the stylar shelves on P'^-M^, although variable in degree of

384

Annals of Carnegie Museum

VOL. 52

development, are more expanded, especially in the parastylar area on
M2-3.

E. acutidens, the youngest recorded species of the genus, has not
been recovered from post-Gardnerbuttean horizons, and apparently
» became extinct sometime during the earliest Bridgerian (Stucky, 1983).
It last appears at one locality (two horizons) in the upper part of the
Lost Cabin Member, where it overlaps in range with Trogosus, and at
Huerfano locality II (McKenna, 1976), where it co-occurs with Tro-
gosus.

Trogosus Leidy, 1871

In his review of Trogosus, Gazin (1953) named two new species {T.
hillsi, T. grangeri) from the Huerfano Basin, and cited five species that
had previously been described from Bridger B horizons in the Bridger
Basin. He recognized two of the latter as valid {T. castoridens, T.
hyracoides), two as invalid {T. minor, T. ? vetulus), and one as ques-
tionable {T. ? latidens), but later (Gazin, 1976) included T. latidens in
his comprehensive faunal list for the Bridger Formation.

Robinson (1966), impressed with the degree of variation in the Tro-
gosus material from Huerfano, suggested that the differences between
T. hillsi and T. grangeri were probably due to sexual dimorphism, and
synonymized the two as T. grangeri.

This is the first record of Trogosus from the Wind River Basin.

Trogosus sp.

(Table 3)

Referred specimens. — Mj or M2, UCM 47717; M^, UCM 46536.

Localities. ~\J CM Iocs. 79040, 81028 (Gardnerbuttean, Lost Cabin
Member).

Known distribution.— Bridgerian— Wind River Basin (Wind
River Fm.), Wyoming; see text for occurrences of Trogosus.

Discussion.— TYiQSQ two teeth closely resemble comparable parts of
the dentition of Trogosus. They are much larger than Esthonyx and
Megalesthonyx, and M^ (Stucky, 1982, Fig. 34) lacks the mesostyle
that is characteristic of the latter. The two teeth are closest in size to
T. castoridens and T. hyracoides, but, in the absence of diagnostic
features of the skull and dentition (Gazin, 1953), the material cannot
be referred to a species of Trogosus. Both of the localities at which
Trogosus occurs are in the upper part of the Lost Cabin Member and
are Gardnerbuttean (= early Bridgerian) in age (see below).

Biostratigraphic Implications

Esthonyx and Trogosus, together with certain other taxa, have prov-
en to be reliable indicators of biostratigraphic age. Esthonyx is common

1983

Stucky and Krishtalka-— Wind River Faunas, 4

385

in most Clarkforkian through Lostcabinian faunas. Its first appearance,
along with that of other taxa, marks the onset of the Clarkforkian (Rose,
1981). The E. xenicus-ancylion-grangeri lineage is characteristic of the
Clarkforkian in the Bighorn Basin (Gingerich and Gunnell, 1979) and
becomes extinct in the early Wasatchian. The E. spatularius-bisulcatus
chronospecies is characteristic of Graybullian and Lysitean horizons,
and apparently extends into the Lostcabinian in the Bighorn Basin
(Schankler, 1980). The first appearance of E. acutidens may mark the
beginning of the Lostcabinian; this species is an index fossil of that
subage, but it is also known from post-Lostcabinian (Gardnerbuttean)
faunas at two localities —one in the Huerfano Formation (Huerfano
II), and one in the Wind River Formation (CM. loc. 34, upper part
of Lost Cabin Member). Its last appearance (contra McKenna, 1976)
is not necessarily indicative of a late Wasatchian age.

Before the Bridgerian Land-Mammal Age was formally named, ho-
rizons in the Huerfano Formation (Huerfano B) that yielded Trogosus,
as well as certain other taxa, were correlated with part of the Bridger
Formation (Osborn, 1897, 1909, 1919, 1929; Matthew, 1899, 1909;
Simpson, 1933). Subsequently, the Wood Committee (Wood et al.,
1941) listed Tillotherium {=Trogosus) as an index fossil of the Bridg-
erian, and assigned this age to the fauna from Huerfano B. Robinson
(1966), however, concluded that the Huerfano B fauna was represen-
tative of a new, latest Wasatchian subage, the Gardnerbuttean, which
was younger than the Lostcabinian but older than the Bridgerian. Later,
a new record of E. acutidens from Huerfano locality II (Huerfano B)
led McKenna (1976) to question the utility of the Gardnerbuttean.
Huerfano locality II is the only fossiliferous horizon that preserves the
co-occurrence of Esthonyx and Trogosus.

Trogosus has also been recorded from the Aycross Formation
(McKenna, 1980; Bown, 1982; CM collections), the Cathedral Bluffs
Tongue of the Wasatch Formation (Morris, 1954; Gazin, 1 962), Bridger
A and B horizons in the Bridger Basin (McGrew and Sullivan, 1971;
Gazin, 1976), the Green River Formation (CM collections. Powder
Wash, Utah), and possibly from the Arkosic facies of the New Fork
Tongue of the Wasatch Formation (Stucky, 1983). The record of Tro-
gosus from the Wind River Basin comes from two localities in the
upper part of the Lost Cabin Member, UCM Iocs. 79040 and 81028.
The former also preserves Palaeosyops {=Eotitanops, see Wallace, 1980)
borealis, Hyrachyus sp. cf. H. eximius, Antiacodon pygmaeus, and Pan-
tolestes sp. cf. P. longicaudus. The M^ of Trogosus is the only known
specimen from UCM loc. 81028.

These X3.x2i — Trogosus, Palaeosyops, Hyrachyus, Antiacodon, Pan-
tolestes— and others, such as Helohyus and Helaletes, are restricted in
the Wind River Formation to localities in the upper part of the Lost

386

Annals of Carnegie Museum

VOL. 52

Cabin Member. All of these taxa appear to be reliable index fossils of
the Bridgerian. Most of them occur in faunas of unequivocal Bridgerian
age (Gazin, 1976; Bown, 1982; McKenna, 1980; McGrewand Sullivan,
1971) and their first appearance can be used to mark the onset of the
Bridgerian (Stucky, 1983). Some faunas— from Huerfano B, Cathedral
Bluffs, upper part of the Lost Cabin Member— in which many of these
Bridgerian taxa co-occur with typical late Wasatchian mammals, such
as Esthonyx, Shoshonius, Coryphodon, Bunophorus, Diacodexis, Hy-
racotherium, and Phenacodus, have been assigned alternatively to either
the latest Wasatchian (Robinson, 1966; Gazin, 1962; McKenna, 1976)
or the earliest Bridgerian (Morris, 1954; West and Dawson, 1973;
Gingerich, 1979; Bown, 1982; Stucky, 1982). We opt for the latter
conclusion on the basis that the first appearances of the Bridgerian taxa,
which apparently occur penecontemporaneously, are more reliable cri-
teria of the Wasatchian-Bridgerian boundary than are the last appear-
ances of the Wasatchian taxa, which are asynchronous and geograph-
ically disjunct. In short, the principle advocated here is that the first
appearance of taxa through immigration or cladogenesis is a finer cor-
relative tool and biostratigraphic measure than is the extinction and
last appearance of taxa, or stage of evolution of taxa within an ana-
genetic lineage (Repenning, 1967; Savage, 1977; Woodbume, 1977;
Rose, 1981).

Accordingly, the faunas from localities in the upper part of the Lost
Cabin Member, as well as those from Huerfano B, Cathedral Bluffs,
and the Arkosic facies of the New Fork Tongue of the Wasatch For-
mation (in part) are here considered Bridgerian. They preserve the first
appearance of the Bridgerian taxa Trogosus, Antiacodon, Palaeosyops,
and Hyrachyus, and the unique, earliest Bridgerian faunal assemblages
that define the Gardnerbuttean (Stucky, 1983).

Phylogenetic Relationships

All North American tillodonts appear to be united in having enlarged
P/2, strong, broad postcingula and stylar shelves on the upper molars,
a talonid basin on P3_4, high arcuate crests and metastylids on the lower
molars, and in the loss of PVi and P (Fig. 3, node 1). Two lineages of
Esthonyx, the most primitive North American tillodont, appear to have
diverged in the early Tertiary. One, the E. xenicus-ancylion-grangeri
chronospecies, retains an unfused mandibular symphysis and a two-
rooted P2, but seems to be derived in having the paraconid on the
lower molars close to an anteriorly inflated metaconid (Fig. 3, node 2).
The second lineage of Esthonyx and all other North American tillo-
donts (Fig. 3, node 3) are advanced in having a fused mandibular
symphysis (to at least below P3), greatly reduced conules on M‘"^, and
more hypsodont and deeply basined P4-M3, with higher, more arcuate

1983

Stucky and Krishtalka— Wind River Faunas, 4

387

Fig. 3. — Proposed phylogenetic relationships among Tillodontia. Node A — with
wing-like stylar salients, and postcingulum. Node B—\2 enlarged; with more ex-
panded stylar shelf and postcingulum. Node 7 —P/2 more enlarged; Mi_2 more hypsodont,
with strong metastylid, arcuate paracristid and posthypocristid, reduced hypoconulid;
M’-^ less transverse; P, P'/i lost; P3_4 with talonid basin (?primitive). Node 2— Mi_3
paraconid closer to anteriorly inflated metaconid; chronocline of increasing size from E.
xenicus-E. ancylion-E. grangeri. Node i— Mandibular symphysis fused to below P3 or
P4; P4-M3 more hypsodont, with higher cristids, deeper basins and more U-shaped
paracristid; M'-^ with reduced conules. Node 4— with expanded stylar salients and
deeper ectoflexi; P2 single-rooted. Node 5— P/2 enlarged, with restricted band of enamel;
P4-M3 more hypsodont; M,_3 cristid obliqua more lingual, originating from below the
metaconid; increase in size; reduction of Ci, P2, P4 talonid. Node 6— with mesostyle.

7— Mandibular symphysis fused to below Mi; P/2 greatly enlarged; increase in size;
longer P-C-P^ diastemata; P2 single-rooted; P/,, C, P3 talonid reduced. Node 7^1 — 13 lost;
longer C--P2 diastema.

cristids and stronger metastylids. E. bisulcatus {=E. spatularius) and
E. acutidens (Fig. 3, node 4) have, in addition to these features, ex-
panded stylar salients and deeper ectoflexi on M^-^, and a single-rooted
P2. E. bisulcatus appears to be one Graybullian through Lysitean chron-
ospecies; the lack of overlap in size between the latter and E. acutidens,
and their apparent co-occurrence in Lostcabinian beds in the Bighorn
Basin may imply specific distinction and a cladogenetic event.

Remaining liWodonXs— Megalesthonyx, Trogosus, Tillodon (Fig. 3,
node 5)— are much larger and more derived in having a restricted band
of enamel on P/2 (P not known in Megalesthonyx), a cristid obliqua
on the lower molars that originates lingually from the metaconid, more
hypsodont cheek teeth, and reduced Cj, P2, and P4 talonid. In Mega-

388

Annals of Carnegie Museum

VOL. 52

lesthonyx (Fig. 3, node 6) a mesostyle is developed on the only known
upper molar. Trogosus and Tillodon (Fig. 3, node 7) are still larger,
with the mandibular symphysis extending to below Mi, a greatly en-
larged P/2, single-rooted P2, longer diastemata between P-C-P^, and
reduced P/,, C, and P3 talonid. Tillodon (Fig. 3, node 7 A) appears to
be most advanced in the loss of P/3 and in the enlarged C1-P2 diastema.

Most non-North American tillodonts are known from relatively poor
material. E. munieri, based on three isolated teeth from the early Eocene
of France, cannot at present be allied with either lineage of Esthonyx
in North America. Basalina, known from a fragmentary dentary with
a broken P4 from the Eocene of Pakistan, has a fused mandibular
symphysis (to below P4), and unreduced Ci, P2 (double rooted), and
P4 talonid (Lucas and Schoch, 1981). As such, it appears to be less
derived than Megalesthonyx, Trogosus, and Tillodon, and more closely
related to E. bisulcatus and E. acutidens (Fig. 3, node 3). Given the
absence of PVj in all known tillodonts (Gazin, 1953), the reconstruction
of the dental formula of Basalina by Lucas and Schoch (1981) in the
diagnosis (p. 90) and their figure (Plate 1 5, Fig. 2) is questionable. More
reasonable is the alternative interpretation (p. 91) of a two-rooted P2
and the absence of Pj.

Lofochaius and Meiostylodon, from the early or middle Paleocene
of China (Li and Ting, 1983), are known from P^-M^ and from M^"^
and I2, respectively. They are tillodont-like in the enlargment of I2 and
in the wing-like stylar salients and expanded postcingula on the cheek
teeth. In comparison to Esthonyx, Lofochaius, with transverse upper
molars and narrower stylar shelves and postcingula, is the most prim-
itive known tillodont (Fig. 3, node A). Meiostylodon retains transverse
M^~2 unreduced conules, but approaches Esthonyx in the devel-
opment of the postcingula and stylar shelves, and appears to be inter-
mediate morphologically between Lofochaius and Esthonyx (Fig. 3,
node B). Esthonyx and all other tillodonts (Fig. 3, node 1) are more
derived, in comparison, in having less transverse P'^-M^. The other
characters listed for this node are not preserved in Lofochaius and
Meiostylodon-, some or all may or may not have been present in these
taxa.

Adapidium, from the late Eocene of China, has hypsodont lower
molars with a lingual cristid obliqua that originates from below the
metaconid (Fig. 3, node 5). It appears to be most closely related to
Megalesthonyx (Fig. 3, node 6) in sharing the development of a me-
sostyle on at least M* (unpublished, unnumbered IVPP specimen),
although this may have evolved independently.

Kuanchuanius (see Chow, 1963), from the middle Eocene of China,
shares with Trogosus (Fig. 3, node 7) the derived features of very large
size, a mandibular symphysis that extends to below Mj, a greatly en-

1983

Stucky and Krishtalka— Wind River Faunas, 4

389

larged I2 with a restricted band of enamel, reduced I, and Cj, a single-
rooted P2, and hypsodont lower molars.

The inferred phylogenetic relationships of the four Chinese Paleogene
tillodonts may suggest that the origin of tillodonts occurred in Asia
during the early Paleocene, and was followed by at least two periods
of origin and immigration of North American taxa. Meiostylodon and
Kuanchuanius are suitable morphologically and temporally to be near
the ancestry of the larger Esthonyx and Trogosus, respectively. Me-
galesthonyx appears abruptly in the Lostcabinian of North America
and may have shared a common ancestry with Adapidium. The pres-
ence of a mesostyle on seems to remove Megalesthonyx from the
ancestry of Trogosus or Tillodon. Less plausible is the alternative hy-
pothesis (Rose, 1972), which involves the appearance and subsequent
loss of the mesostyle in an Esthonyx- Megalesthonyx-Trogosus lineage.

Acknowledgments

We thank M. R. Dawson for reviewing the manuscript, P. Gingerich for his criticisms
and information on Bighorn Basin tillodonts, and V. Abromitis for his photographic
assistance. This work was supported, in part, by grants from the M. Graham Netting
Research Fund of the Carnegie Museum of Natural History through a grant from the
Cordelia S. May Charitable Trust, the Rea Postdoctoral Fellowship, Carnegie Museum
of Natural History, and the Walker Van Riper Fund, University of Colorado Museum.

Literature Cited

Bown, T. M. 1979. Geology and mammalian paleontology of the Sand Creek facies.
Lower Willwood Formation (Lower Eocene), Washakie County, Wyoming. Mem.
Geol. Survey Wyoming, 2:1-151.

. 1982. Geology, paleontology, and correlation of Eocene volcaniclastic rocks,

southeast Absaroka Range, Hot Springs County, Wyoming. Geol. Survey Prof. Pa-
pers, 1201 -A: 1-75.

Chow, M. 1963. Tillodont materials from Eocene of Shantung and Honan. Vert.
Palasiatica, 7:97-104.

Chow, M., Y. Chang, B. Wang, and S. Ting. 1973. New mammalian genera and
species from the Paleocene of Nanhsiung, N. Kwangtung. Vert. Palasiatica, 1 1:31-
35.

Cope, E. D. 1874. Report upon vertebrate fossils discovered in New Mexico, with
description of new species. Annual Rept. Chief Engineers, U.S. Army, Appendix
FF, p. 589-606 (Separatum, pp. 1-18).

. 1880. The northern Wasatch fauna. Amer. Nat., 14:908-909.

. 1881. On the Vertebrata of the Wind River Eocene beds of Wyoming. Bull.

U.S. Geol. Geog. Surv. Terr., 6:183-202.

. 1 883. On the mutual relations of the bunotherian Mammalia. Proc. Acad. Nat.

Sci., Philadelphia, pp. 77-83.

Dehm, R., and T. Oettingen-Spielberg. 1958. Palaontologische und geologische Un-
tersuchungen im Tertiar von Pakistan. 2. Die mitteleocanen Saugetiere von Ganda
Kas bei Basal in Nordwest-Pakistan. Abh. Bayer. Akad. Wiss., Math-Nat. Kl., 91:
1-54.

Gazin, C. L. 1953. The Tillodontia: an early Tertiary order of mammals. Smithsonian
Misc. Coll., 121:1-110.

390

Annals of Carnegie Museum

VOL. 52

. 1962. A further study of the Lower Eocene mammal faunas of southwestern

Wyoming. Smithsonian Misc. Coll., 144:l--98.

. 1976. Mammalian faunal zones of the Bridger middle Eocene. Smithsonian

Contrib. Paleobiol., 26:1-25.

Gingerich, P. D. 1979. Phylogeny of middle Eocene Adapidae (Mammalia, Primates)
in North America: Smilodectes and Notharctus. J. Paleont., 53:153-163.

Gingerich, P. D., and G. F. Gunnell. 1979. Systematics and evolution of the genus
Esthonyx (Mammalia, Tillodontia) in the early Eocene of North America. Contribs.
Mus. Paleont., Univ. Michigan, 25:125-153.

Guthrie, D. A. 1967. The mammalian fauna of the Lysite Member, Wind River
Formation, (early Eocene) of Wyoming. Mem. So. California Acad. Sci., 5:1-53.

. 1971. The mammalian fauna of the Lost Cabin Member, Wind River For-
mation (Lower Eocene) of Wyoming. Ann. Carnegie Mus., 43:47-113.

Kelley, D. R., and A. E. Wood. 1954. The Eocene mammals from the Lysite Member,
Wind River Formation of Wyoming. J. Paleont., 28:337-366.

Krishtalka, L., and R. K. Stucky. 1983. Revision of the Wind River faunas, early
Eocene of central Wyoming. Part 3. Marsupialia. Ann. Carnegie Mus., 52:205-227.

Leidy, j. 1968. Notice of some remains of extinct pachyderms. Proc. Nat. Acad. Sci.,
Philadelphia, 20:230-233.

. 1871. Remarks on fossil vertebrates from Wyoming. Proc. Nat. Acad. Sci.,

Philadelphia, 23:228-229.

Lemoine, V. 1889. Considerations generales sur les vertebres fossiles des environs de
Reims et specialement sur les mammiferes de la faune Cemaysienne. C. R. Congr.
Intemat. Zool., p. 233-279.

Li, Chuan-Kuei, and Su-Yin Ting. 1983. The Paleogene mammals of China. Bull.
Carnegie Mus. Nat. Hist., 21:1-98.

Lucas, S. G., and R. M. Schoch. 1981. Basalina, a tillodont from the Eocene of
Pakistan. Mitt. Bayer. Staatslg. Palaont. hist. Geol., 21:89-95.

Lucas, S. G., R. M. Schoch, E. Manning, and C. Tsentas. 1981. The Eocene bio-
stratigraphy of New Mexico. Bull. Geol. Soc. Amer., pt. 1, 92:951-967.

Marsh, O. C. 1875. New order of Eocene mammals. Amer. Jour. Sci., 9:221.

Matthew, W. D. 1899. A provisional classification of the freshwater Tertiary of the
west. Bull. Amer. Mus. Nat. Hist., 12:19-75.

. 1 909. Faunal lists of the Tertiary Mammalia of the west. Bull. U.S. Geol. Surv.,

361:91-138.

McGrew, P. O., and R. Sullivan. 1970. The stratigraphy and paleontology of Bridger
A. Univ. Wyoming. Contribs. Geol., 9:66-85.

McKenna, M. C. 1976. Esthonyx in the upper faunal assemblage, Huerfano Formation,
Eocene of Colorado. J. Paleont., 50:354-355.

. 1980. Late Cretaceous and early Tertiary vertebrate paleontological recon-
naissance, Togwotee Pass area, northwestern Wyoming. Pp. 321-343, in Aspects of
vertebrate history (L. L. Jacobs, ed.). Museum of Northern Arizona Press, Flagstaff,
xix + 407 pp.

Morris, W. J. 1954. An Eocene fauna from the Cathedral Bluffs Tongue of the Washakie
Basin, Wyoming. J. Paleont. 28:195-203.

Osborn, H. F. 1 897. The Huerfano Lake Basin, southern Colorado and its Wind River
and Bridger fauna. Bull. Amer. Mus. Nat. Hist., 9:247-258.

. 1909. Cenozoic mammal horizons of western North America, with faunal lists

of the Tertiary mammals of the west by William Diller Matthew. Bull. U.S. Geol.
Surv., 361:1-138.

. 1919. New titanotheres of the Huerfano. Bull. Amer. Mus. Nat. Hist., 41:557-

569.

. 1 929. The titanotheres of ancient Wyoming, Dakota, and Nebraska. U.S. Geol.

Surv. Monograph 55.

1983

Stucky and Krishtalka—Wind River Faunas, 4

391

Repenning, C. A. 1 967. Palearctic-Nearctic mammalian dispersal in the late Cenozoic.
Pp. 288-3 \\,in The Bering Land Bridge (D. M. Hopkins, ed.), Stanford Univ. Press,
Stanford, California.

Robinson, P. 1 966. Fossil Mammalia of the Huerfano Formation, Eocene of Colorado.
Bull. Yale Peabody Mus. Nat. Hist., 21:1-95.

Rose, K. D. 1972. A new tillodont from the Eocene upper Willwood Formation of
Wyoming. Postilla, 155:1-13.

. 1981. The Clarkforkian Land-Mammal Age and mammalian faunal compo-
sition across the Paleocene-Eocene boundary. Univ. Michigan Papers Paleont., 26:
1-197.

Savage, D. E. 1977. Aspects of vertebrate paleontological stratigraphy and geochron-
ology. Pp. 427-442, in Concepts and methods of biostratigraphy (E. G. Kaufmann
and J. E. Hazel, eds.), Dowden, Hutchinson and Ross, Stroudsburg, Pennsylvania.

ScHANKLER, D. M. 1980. Faunal zonation of the Willwood Formation in the central
Bighorn Basin, Wyoming. Pp. 99-1 14, in Early Cenozoic paleontology and stratig-
raphy of the Bighorn Basin, Wyoming (P. D. Gingerich, ed.), Univ. Michigan Papers
Paleont., 24: vi + 1-146.

Simpson, G. G. 1933. Glossary and correlation charts of North American Tertiary
mammal-bearing formations. Bull. Amer. Mus. Nat, Hist., 60:79-121.

. 1937. Notes on the Clark Fork, upper Paleocene fauna. Amer. Mus. Novit.,

954:1-24.

Simpson, G. G., A. Roe, and R. C. Lewontin. 1960. Quantitative Zoology. Revised
ed., Harcourt, Brace and Co., New York.

SoKAL, R. R., AND F. J. Rohlf. 1 969. Biometry. W. H. Freeman and Co., San Francisco.

Stucky, R. K. 1982. Mammalian faunas and biostratigraphy of the upper part of the
Wind River Formation (early to middle Eocene), Natrona County, Wyoming, and
the Wasatchian-Bridgerian boundary. Unpublished Ph.D. dissertation, Univ. Col-
orado, Boulder, 278 pp.

. 1983. Geology and mammalian biostratigraphy of the upper part of the Wind

River Formation (early to middle Eocene) of Wyoming, and the Wasatchian-Bridg-
erian boundary. Bull. Carnegie Mus. Nat. Hist., in press.

Stucky, R. K., and L. Krishtalka. 1982. Revision of the Wind River Faunas, early
Eocene of central Wyoming. Part 1. Introduction and Multituberculata. Ann. Car-
negie Mus., 51:39-56.

Wallace, S. M. 1980. A revision of North American early Eocene Brontotheriidae
(Mammalia, Perissodactyla). Unpublished M.S. thesis, Univ. Colorado, Boulder,
157 pp.

Wang, B. 1975. Paleocene mammals of Chaling Basin, Hunan. Vert. Palasiatica, 13:
154-162. [Chinese]

West, R. M., and M. R. Dawson. 1973. Fossil mammals from the upper part of the
Cathedral Bluffs Tongue of the Wasatch Formation (early Bridgerian), northern
Green River Basin, Wyoming. Univ. Wyoming Contrib. Geol. 12:33-41.

White, T. E. 1952. Preliminary analysis of the vertebrate fauna of the Boysen Reservoir
area. Proc. U.S. Nat. Mus., 102:185-207.

Wood, A. E., and others. 1941. Nomenclature and correlation of the North American
continental Tertiary. Bull. Geol. Soc. Amer., 52(1): 1-48.

WooDBURNE, M. O. 1977. Definition and characterization in mammalian chronostra-
tigraphy. J. Paleont., 51:220-234.

Young, C. C. 1 937. An early Tertiary vertebrate fauna from Yuanchu. Bull. Geol. Soc.
China, 17:413-438.

Back issues of many Annals of Carnegie Museum articles are
available, and a few early complete volumes and parts are listed
at half price. Orders and inquiries should be addressed to:
Publications Secretary, Carnegie Museum, 4400 Forbes Avenue,
Pittsburgh, Pa. 15213.

/

? 1. "

t.:.

CViVisi

'^kt 'b

. ., "

...Ml-'" „
r-i.t,;H J. ■

%: t. 1 ‘•*‘-,.i

i\tif -.y*-

,l - '

■! ^<^|«*vSfs fr(

t

ff

■s'.+'f*^’*^* - >it' ,., ! I “^ ' ;. 1; .■ ■

. ' >'' -i ■ -■ <■"

r'f,:~ -V. ■' •■' ■■ . ^

• ■ iX ■ j'jfS' ,! /“ T ■ ■ . •,’'*, ■>-'.» <1 i

• is;-,'' >■ ; ■-■rs ‘

‘’•' !*:•'' , ‘ /••

• < ^ ul V* ♦ M J '■ ■ > vf:>>

4 ' .

Ia ^;l ^!' ■■' , “

B,. V . .. ' ' Itrr 4

mJ. ^.■,,v - ''"'*-‘1# ■ ; i,r l?, ''•■■. •■■

. A-.,'-; , ■ ' “ V ^

>aiit

taK *!'Vi'

'J'

', '■; I

*t.'5 V. - ■ r'"

-; ^ .■ u -, .-

!**•*

•J ^

■iS ■ ^

>.:■

7. 73

ISSN 0097-4463

AN N ALS

of CARNEGIE MUSEUM

CARNEGIE MUSEUM OF NATURAL HISTORY

4400 FORBES AVENUE « PITTSBURGH, PENNSYLVANIA 15213

VOLUME 52 7 DECEMBER 1983 ARTICLE 18

CAPTORHINOMORPH “STEM” REPTILES FROM THE
PENNSYLVANIAN COAL-SWAMP DEPOSIT OF
LINTON, OHIO

Robert R. Reisz^

Research Associate, Section of Vertebrate

Donald Baird^

Abstract

Two new specimens, plus one previously misidentified as an adelospondylous am-
phibian bring to four the known captorhinomorph reptile remains in the classic Linton
fauna. The skull and jaw specimens are clearly assignable to the Mazon Creek genus
Cephalerpeton, although specifically distinct from C ventriarmatum-, the postcranial
specimen conforms to the Linton species Anthracodwmeus longipes. As the question of
possible synonymy between these two nominal genera cannot be resolved on the evidence
available, both names are retained sub judice. The Linton species designated Cephal-
erpeton aff. C ventriarmatum (of late Westphalian D age) may have been derived from
the Mazon Creek species (early Westphalian D) through enlargement of the mandibular
teeth and resultant reduction of the dental formula. These agile, lizard-like small reptiles
occur as rare erratics in the Linton deposit.

Introduction

The suborder Captorhinomorpha occupies a unique position in the
early evolution of reptiles. The fossil record of this group extends from
the Lower Pennsylvanian into the Upper Permian, forming one of the
longest reptilian phylogenies within the Paleozoic. The Captorhino-
morpha include the oldest known true reptiles (Eureptilia) and are

' Address: Department of Biology, University of Toronto, Erindale Campus, Mississauga,
Ontario, L5L 1C6, Canada.

2 Address: Museum of Natural History, Princeton University, Princeton, New Jersey,
08544.

Submitted 24 January 1983.

393

394

Annals of Carnegie Museum

VOL. 52

closely related to the important synapsid (pelycosaur and therapsid)
and diapsid (eosuchian, lepidosaur, and archosaur) lineages. The sub-
order can be divided into two families, the Protorothyrididae and the
Captorhinidae. The protorothyridid genera were formerly included in
the family Romeriidae (Carroll and Baird, 1972), but with the removal
of the type genus Romeria to the Captorhinidae (Heaton, 1979) the
family name Protorothyrididae Price, 1937 (emend. Gregory, 1950)
becomes the correct designation for this group (Reisz, 1980).

Members of the more advanced captorhinomorph family, the Cap-
torhinidae, occur in large numbers in terrestrial strata of Early to Late
Permian age from the equatorial Laurasian land mass (Olson, 1952,
1954, 1962; Olson and Beerbower, 1953; Konzhukova, 1956; Heaton,
1979) and have also been found recently on the Gondwana land mass
(Taquet, 1969; Kutty, 1972; Gaffney and McKenna, 1979). The cap-
torhinids form a compact group of structurally similar, relatively spe-
cialized, slow, heavy-set reptiles. The structural changes that occur in
this family include an approximately hve-fold increase in size during
the Permian and an increase in the numbers of rows of maxillary and
dentary teeth.

The fossil record of the more primitive family, the Protorothyrididae,
is restricted to a few specimens from the Pennsylvanian and Lower
Permian deposits on the North American and European land masses
of Laurasia. The known members of this family also form a compact
group of structurally similar animals, but, in strong contrast to the
captorhinids, the protorothyridids are relatively generalized, small, ag-
ile reptiles. Our current understanding of the Protorothyrididae is based
primarily on the studies by Carroll (1964, 1969), Carroll and Baird
(1972), and Clark and Carroll (1973).

The Pennsylvanian fossil record of this phylogenetically important
group of reptiles appears to be restricted in variety and numbers by
their preference for dry ground, away from the environments typically
preserved during this period. Only as a result of preservation under
unusual circumstances are we able to study a few remains of early
protorothyridids. Specimens of Hylonomus lyelli and Paleothyris aca-
diana have been recovered from inside Sigillaria tree stumps that were
preserved in standing position near Joggins and Florence, Nova Scotia,
respectively (Carroll, 1964, 1969). A single skeleton of Cephalerpeton
ventriarmatum has been found in an ironstone nodule from Mazon
Creek, Illinois (Gregory, 1948, 1950; Carroll and Baird, 1972). Single,
nearly complete skeletons of Coelostegus prothales and Brouffia ori-
entalis from Nyfany, Czechoslovakia, and a poorly preserved, im-
mature skeleton of Anthracodromeus longipes from Linton, Ohio, con-
stitute the only protorothyridid material hitherto known from these
famous coal-swamp deposits (Carroll and Baird, 1972). Although the
coal-swamp faunas of Linton and Nyfany have been studied for more

1983

Reisz and Baird— Captorhinomorph Reptile Remains

395

than a century, and thousands of specimens representing scores of fish
and amphibian genera have been collected, only seven specimens of
fully terrestrial reptiles have been found. These animals were probably
erratics rather than customary members of the aquatic communities
with which their remains are associated. The ecologies of the West-
phalian coal-swamp deposits in which reptiles occur have been dis-
cussed by Westoll (1944), Rayner (1971), and most recently by Milner
(1980). A revised list of the tetrapods present at Linton has been pub-
lished by Hook (1981).

The purpose of this paper is to describe two additional specimens
of protorothyridid reptiles that were recently collected at the Linton
mine dump by Dr. Richard Lund, together with a previously misiden-
tified specimen in the British Museum collection. Our study was based
on high-fidelity latex casts (Baird, 1955) of skeletons preserved as nat-
ural molds in carbonaceous shale (Linton) or sideritic mudstone (Ma-
zon Creek).

Abbreviations.— XMNH, American Museum of Natural History; BM(NH), British
Museum (Natural History); CM, Carnegie Museum of Natural History; YPM, Peabody
Museum of Natural History, Yale University.

Key to abbreviations used in the figures:

a = astragalus

p for = parapineal foramen

ang = angular

pmx = premaxilla

c = calcaneum

po = postorbital

ch = chevron

pp = postparietal

cr = caudal rib

prf = prefrontal

d = dentary

prv = presacral vertebra

f = frontal

ptf = posttemporal fenestra

fib = fibula

q = quadrate

il = ilium

qj = quadratojugal

is = ischium

sa = surangular

j = jugal

scl = sclerotic plate

1 = lacrimal

sp = splenial

mt = metatarsal

sq = squamosal

mx = maxilla

St = supratemporal

p = parietal

t == tabular

pf = postfrontal

tib = tibia

V = vomer

Systematic Paleontology

Class Reptilia Linnaeus, 1758
Subclass Eureptilia Olson, 1947
Order Captorhina Olson, 1947
Suborder Captorhinomorpha Watson, 1917
Family Protorothyrididae Price, 1937
Cephalerpeton Moodie, 1912

Figured specimens.— CM 23055 (Fig. 1), a crushed, distorted skull;
the left premaxilla, maxillae and lacrimals, right frontal and parietal.

396

Annals of Carnegie Museum

VOL. 52

left prefrontal, right vomer, and a sclerotic plate are identifiable. The
mandibles are represented by the dentaries, left surangular, and possibly
right splenial. Other skull bones are present but are too badly frag-
mented for precise identification.

BM(NH) R.2667 (Fig. 2), the anterior part of a right mandible in
lingual aspect (purchased as part of the J.W. Davies Collection, 1895).
This specimen was mentioned by Steen (1931:885) as the “lower jaw
of an Adelospondyl.”

Description

Skull — ThQ identifiable portion of the skull CM 23055 resembles
the type specimen of Cephalerpeton ventriarmatum (YPM 796) in a
number of significant features. As in the type specimen, the maxillary
process of the premaxilla is short, with places for only three teeth.
Although Carroll and Baird (1972) reconstructed YPM 796 with a long
maxillary process of the premaxilla, as is the case in other primitive
captorhinomorphs, restudy of the premaxilla reveals that only three
teeth can be accommodated on this bone. A three-toothed premaxilla

1983

Reisz and Baird— Captorhinomorph Reptile Remains

397

Icm

Fig. 2. — Cephalerpeton afF. C. ventriarmatum, BM(NH) R.2667, right dentary exposed
in medial view.

was shown in earlier reconstructions by Gregory (1948, Fig. 2) and
Baird (1965, Fig. 6).

The fact that the maxillae are exposed in lateral view in CM 23055,
but in partial medial view in YPM 796, makes direct comparisons
difficult. In CM 23055 the maxilla, completely separated from the other
skull elements, has a large dorsal expansion above the caniniform teeth.
The dorsally directed plate of bone is large enough to have extended
dorsally to meet the nasal and to have covered the anterior process of
the lacrimal that may have extended to the narial opening. In the type
specimen of Cephalerpeton ventriarmatum the maxillae and lacrimals,
exposed in medial view, have remained in articulation so that the
external relationship of these bones cannot be established.

The similarity between the left lacrimals in the two specimens, both
exposed in medial view, is striking. In YPM 796 the lacrimal duct
opens to the interior near the mediodorsal margin of the maxilla and
just anterior to the midpoint between the orbit and the external naris.
The open groove for the lacrimal duct extends anteroventrally and its
ventral margin forms a narrow ridge above the mediodorsal margin of
the maxilla. The posterior part of the lacrimal has a long suborbital
flange that may have excluded the maxilla from the orbit; much of the

398

Annals of Carnegie Museum

VOL. 52

orbital margin of the lacrimal is covered medially by a long ventral
process of the prefrontal. The same condition occurs in CM 23055. In
this specimen the lacrimal duct is only partly covered by bone. A
striated region on the medial surface of the lacrimal just ventral to the
open region of the lacrimal duct indicates that this area was covered
by a medial shelf of the maxilla to give the same arrangement as in
Cephalerpeton ventriarmatum. A long groove in CM 23055 extends
ventrally from the dorsal tip of the orbital margin to the level of the
lacrimal ridge; this groove held the ventral process of the prefrontal.
The suborbital flange of the lacrimal is also well developed.

The other elements of the skull roof in CM 23055 are not readily
comparable with those in the type specimen of Cephalerpeton ventriar-
matum. In CM 23055 these elements (prefrontal, frontal and parietal)
are preserved so as to expose their external surfaces, whereas the same
bones in YPM 796 are preserved with their medial surfaces exposed.

The roofing elements have a relatively well developed pattern of
sculpturing of low ridges and small pits. The prefrontal has part of its
anterior and posterior dorsal processes exposed, in addition to the long
ventral process which forms part of the anterior orbital margin. The
frontal has a pair of well developed, long grooves along its lateral margin
for the attachment of the prefrontal and postfrontal. Between these
grooves the frontal has a short, laterally directed process that forms
part of the dorsal orbital margin. The medial margin of the badly
crushed right parietal is overlapped by the frontal; its features are
analyzed in detail in a subsequent section. Overlapping the postero-
lateral comer of the parietal is a sclerotic plate that is similar to those
preserved in YPM 796; it is a thin, subrectangular bone with a bevelled
edge.

Mandible. ~F2LYts of the mandibles preserved in CM 23055 include
both dentaries and fragments of the right surangular and splenial. Al-
though both dentaries are partially covered by surrounding elements,
sufficiently large portions of the left medial and right lateral surfaces
are exposed to describe this element. The dentary is unusually deep
dorsoventrally throughout its length. In association with this unusual
depth, the anterior portion of the dentary near the symphysis is some-
what more massive than in other protorothyridids of similar size; more
significantly, the dentary is twice as deep dorsoventrally at the level of
the twelfth tooth than at the third. Although all skull elements are
severely crushed, preservation cannot account for the unusual propor-
tions of the dentary. The dorsal edge of the dentary is concave in lateral
view, providing additional evidence of the unusual depth of the man-
dible in the posterior half of the dentary, and probably in the rest of
the mandible. The alveolar shelf is also unusually large, as exposed on
the left dentary, and its sutural surface with the splenial is extensive.

1983

Reisz and Baird— Captorhinomorph Reptile Remains

399

Only the left dentary is sufficiently exposed to indicate that it had places
for 1 7 or 1 8 teeth. The thirteenth tooth of this dentary has been hgured
by Currie (1979, Fig. 13b), greatly magnified, to illustrate the pattern
of vertical lingual striations on teeth of most Paleozoic reptiles. A nearly
complete right dentary is exposed in medial view in BM(NH) R.2667
(Fig. 2). This element is identical to the left dentary of CM 23055,
showing the typically deep subalveolar expanse of bone, and the large
teeth with infolding of the enamel. As in CM 23055, there is only place
on the dentary of BM(NH) R.2667 for seventeen teeth. Unfortunately
the symphysis is not preserved.

The right surangular is partially exposed between the dentaries of
CM 23055. Gentle striae on its surface radiate from the slightly grooved
dorsal margin; the groove probably represents the area where the cor-
onoid attached. The identification of another fragment lying between
the dentaries as the right splenial is tentative. Longitudinal striations
near its dorsal edge correspond to the expected sutured area with the
alveolar shelf of the dentary.

Dentition. —The most significant similarity between CM 23055 and
YPM 796 is their dentition. In both specimens there are places for
three premaxillary and 1 6 maxillary teeth. This is far below the number
in other primitive captorhinomorphs—Z/j^/owomw^ has a dental for-
mula of 5 + 36, Paleothyris has 6 + 35, and Protorthyris has 5 + 30
(Carroll and Baird, 1972; Clark and Carroll, 1973). In both specimens
the first premaxillary tooth is large, roughly equal to the caniniform
tooth on the maxilla, and much larger than in other primitive captor-
hinomorphs. The larger teeth show some infolding of the enamel. The
palatal dentition is also quite similar in the two specimens, with closely
packed small denticles covering most of the vomers.

Discussion

Similarities between CM 23055 and BM(NH) R.2667 from the Mid-
dle Pennsylvanian (late Westphalian D) deposit at Linton, Ohio, and
YPM 796 from a slightly older horizon (early Westphalian D) at Mazon
Creek, Illinois, indicate that these reptiles can be placed in the same
genus. There are, however, some significant features that distinguish
the Linton specimens from Cephalerpeton ventriarmatum. The pre-
served portion of the dentaries in CM 23055 and BM(NH) R.2667
demonstrates that the lower jaw was deeper dorsoventrally than the
corresponding portions of the lower jaw in C. ventriarmatum. The teeth
on the lower jaws of the Linton specimens are also larger and fewer in
number than those of C. ventriarmatum. In the latter species the teeth
on the dentaries are not substantially different from those in other
protorothyridids, but there are places for only 21 or 22 teeth on the
dentary, a much lower number than in other genera (Carroll and Baird,

400

Annals of Carnegie Museum

VOL. 52

1972). The dentaries of the skull from Linton (CM 23055) have places
for no more than 1 8 large teeth (the exact number is uncertain), whereas
the British Museum dentary has 1 7 (with possibly one more concealed
at the posterior end). In these features the Linton specimen appears
even more advanced than C ventriarmatum. These differences are
compatible with a postulated ancestor-descendant relationship in which
the geologically older species is understandably more primitive in cer-
tain osteological features.

Anthmcodromeus Carroll and Baird, 1972

Figured specimen. — CM 25282 (Fig. 3), consisting of one dorsal and
ten caudal vertebrae, caudal ribs, chevron bone, scattered ventral scales,
and parts of the pelvic girdle and hind limbs.

Description

CM 25282 includes a single posterior dorsal vertebra which has the
general proportions characteristic of all protorothyridid captorhino-
morphs. The centrum is a relatively long, low cylinder, pinched in at
the middle. The neural arch is not swollen, and the neural spine is
relatively tall and blade-like. The shape of the neural spine is remark-
ably similar to those of the dorsal vertebrae in the type specimen of
Anthracodromeus longipes (AMNH 6940). As shown by Carroll and
Baird (1972), neural spines of the dorsal vertebrae are hatchet-shaped,
expanding anteroposteriorly from normal-sized bases to wide summits.
This feature of the neural spines distinguishes Anthracodromeus from
all other protorothyridids.

The lateral surfaces of the neural spines in the type specimen of
Anthracodromeus longipes have a peculiarly “sculptured” or “ham-
mered” appearance. In our opinion this very unusual feature can be
accounted for by the poor ossification and manner of preservation of
the type skeleton (which is obviously that of a very immature animal).
In such a juvenile individual the perichondral bone sheathing the neural
spines would be so thin that, when subjected to severe compression
(as is typical in Linton material), it would become imprinted with the
spongy texture of the interior. CM 25282 is nearly twice as large as the
type specimen and is well ossified, so crushing during preservation has
not produced the same appearance on its neural spines.

Seven anterior caudal vertebrae are preserved in nearly perfect ar-
ticulation. Although they are incomplete, significant osteological fea-
tures can be determined: The ratio of the length to posterior height of
the centrum is almost 2:1. The centra have diminished in diameter,
indicating that the tail was probably long and slender. The neural arches
are long, narrow structures with slender zygapophyses and relatively
tall neural spines. The best preserved neural spine, seen on the sixth

1983

Reisz and Baird— Captorhinomorph Reptile Remains

401

Fig. 3.—Anthmcodromeus longipes, CM 25282, scattered vertebrae, pelves, hind limbs,
and scattered ventral scales.

vertebra of the series, has the same height as that of the isolated dorsal
vertebra, but its profile tapers upward so that its summit is less than
half as expanded as that of the dorsal vertebra. The three most anterior
vertebrae of the series have well developed transverse processes for the
posteriorly curved caudal ribs. The two most posterior vertebrae have
short, small lateral stumps on the centra which probably represent
transverse processes. Three isolated caudal vertebrae are probably from
the mid-portion of the tail; their centra are still quite long, but there

402

Annals of Carnegie Museum

VOL. 52

are no remnants of transverse proeesses and the neural spines are very
short.

All three elements of the left pelvis are partially exposed in lateral
view; none of their sutures is visible. The only distinctive feature of
this pelvis is the shape of the iliac blade, which expands posterodorsally
above the acetabulum. The medial surface of the right iliac blade, also
exposed in CM 25282, shows a series of grooves near its top that are
probably for the attachment of epaxial musculature. In all these features
this pelvis is similar to that of Coelostegus pwthales Carroll and Baird
(1972) from the coal-swamp deposits of Nyfany, Czechoslovakia.

The ilia in the type specimen of Anthracodromeus longipes appear
to be quite different from that of CM 25282. As described and illus-
trated by Carroll and Baird (1972), they consist of a small acetabular
portion joined to a long, narrow iliac blade. A re-examination of the
specimen, however, indicates that what was interpreted as the entire
blade is only its stout posterior ramus; anterior to this a thinner dorsal
flange can be made out by collating the part and counterpart of the
specimen. As re-interpreted, the iliac configuration appears to be com-
patible with that of CM 25282, if allowance is made for the great
ontogenetic difference between the two individuals; its anteroposterior
width is sufficient to accommodate two sacral ribs. This is at variance
with Carroll and Baird’s interpretation of the sacral region; their re-
construction of the ilium with a long, narrow, posterodorsally oriented
blade could only accommodate one sacral rib. Two sacral ribs are,
however, present in the type of Anthracodromeus longipes, giving added
support to the above interpretation. We must emphasize the difficulty
of interpreting an area that is feebly ossified, squashed flat against the
vertebrae and femora, and obscured by pyritization.

All the limb elements are strongly crushed and distorted, making
most morphological comparisons impossible. The femur appears to be
long and slender. The tibia and fibula are considerably shorter than
the femur, only about 56% as long; this proportion is characteristic of
protorothyridids. The tibia has a wide proximal head and a slightly
narrower distal end. The proximal end of the fibula appears narrow,
whereas the distal end is broad and blade-like. The calcaneum, as in
Paleothyris, (Carroll, 1969) is distinctive in having a distolateral notch.
This notch, not found in other primitive reptiles, separates the distal
surface of articulation with the fourth and fifth distal tarsals from the
lateral edge of the calcaneum. The astragalus is partially obscured by
the tibia, but it was undoubtedly L-shaped. The size of the distal end
of the fibula indicates that the proximal fibular articulation of the
astragalus was smaller than the tibio-astragalar articulation and that
the neck of the astragalus was narrow, as in Paleothyris. The metatarsals

1983

Reisz and Baird-— Captorhinomorph Reptile Remains

403

are long and slender elements, the third metatarsal being nearly equal
in length to the fibula, as in other protorothyridids.

Comparisons

The total eureptilian sample known from the Linton mine deposit,
near Wellsville, Ohio, consists of seven specimens, three of which are
pelycosaurian (Reisz, 1975) and four protorothyridid; all are fragmen-
tary, and only the type of Anthracodromeus longipes includes the major
part of the (very juvenile) skeleton. With such a small sample and such
incomplete specimens it is difficult to judge the number of protoro-
thyridid taxa present, either on anatomical evidence or on the basis of
probabilities.

Once the apparent difference between ilia has been resolved, CM
25282 is reasonably assignable to ''Sauropleura'" longipes Cope, a species
that was tentatively transferred to Cephalerpeton by Baird (1958) and
was subsequently made the type species of Anthracodromeus by Carroll
and Baird (1972). The skull specimen from Linton (CM 23055) and
the isolated mandible (BM[NH] R.2667) are clearly assignable to the
Mazon Creek genus Cephalerpeton although they differ significantly
from the species C ventriarmatum. The question thus arises: are there
actually two protorothyridid genera present in the Linton fauna, or are
Cephalerpeton and Anthracodromeus to be synonymized on the basis
of the new evidence? To answer this requires a further examination of
such anatomical features, both cranial and postcranial, as can be com-
pared in the specimens at hand.

Skull.— As shown in Fig. 4, the type skull of Anthracodromeus lon-
gipes is truncated by the edge of the slab. During compression the cheek
(with which the mandible remained articulated), was folded under the
skull table, and the two sheets of delicate bone were compressed to-
gether so forcibly that the rim of the parapineal foramen embossed a
circle on the jugal. The only skull element that is present in both this
specimen and the new skull of Cephalerpeton from Linton is the right
parietal.

In the latter specimen (Fig. 1), the disarticulated parietal lies with
its dorsal surface uppermost and its anteroposterior axis pointing to-
ward five o’clock; its medial margin is overlapped by the frontal so
that only part of the heavily-rimmed parapineal foramen is exposed.
The posterior margin of the parietal, which faces the prefrontal as the
specimen lies, has its sculptured surface rabbeted by two semicircular
embayments for the articulation of overlapping elements. These rabbets
are floored by a posterior continuation of the parietal, as is indicated
by the way the striations radiate from the center of ossification. This
posterior configuration of the parietal is similar to that seen in An-

404

Annals of Carnegie Museum

VOL. 52

Fig. 4.--Anthracodwmeus longipes, AMNH 6940, partial skull from holotype skeleton,
part and counterpart.

thracodromeus longipes (Fig. 4), where the more lateral embayment
accommodates the supratemporal, while the more medial one, which
is doubly curved, accommodates both the postparietal and the dorsal
process of the tabular. On the anterior margin of the parietal, in CM

1983

Reisz and Baird— Captorhinomorph Reptile Remains

405

23055, are shallow articular facets for the overlapping frontal and post-
frontal, and on the anterolateral comer is one for the postorbital. The
long parietal lappet seen in A. longipes is not in evidence in CM 23055,
but this area is concealed by a sclerotic plate.

The parietal of CM 23055 has all the landmarks of the corresponding
elements in the protorothyridid genera Hylonomus, Brouffia, Coelos-
tegus and Paleothyris, although it differs significantly from them in
proportions, being conspicuously short relative to its width. In the type
of C. ventriarmatum the shape of the parietal has had to be recon-
structed from the configuration of the surrounding elements (Carroll
and Baird 1972, Fig. 2). The restoration shows it as broader than it is
long; but a reconsideration of the palate indicates that the skull has
been made too wide in this region. Thus the true proportions of the
parietal in C. ventriarmatum are too conjectural for valid comparisons
to be made. Similarly, in the type of A. longipes the parietal has been
reconstructed as broad and short (Carroll and Baird 1972, Fig. 5); but
as the anterior part of the bone is missing, its actual proportions remain
uncertain. Thus the parietal, the only cranial element that permits the
two skulls from Linton to be compared with the one from Mazon Creek,
is too incompletely preserved to provide a sound basis for taxonomic
judgement.

The cheek region is similar in the type specimens of C. ventriar-
matum and A. longipes. It differs from the cheeks of other Pennsyl-
vanian protorothyridids in being short anteroposteriorly. This short-
ening of the cheek would be consistent with a shortening of the skull
table (for which, as noted above, direct evidence is lacking).

Mandible.— The only part of the mandible that can be compared
directly in CM 23055 and the type of A. longipes is the posterodorsal
area of the surangular. In both these specimens the mandible appears
to be slightly deeper than in other protorothyridids. Given the incom-
plete nature of the cranial remains of these specimens little reliance
can be placed on small proportional differences. In the type of Ceph-
alerpeton ventriarmatum the lower jaw does not appear to be deeper
than in other protorothyridids; this condition corresponds to the rel-
atively unspecialized mandibular dentition present in this species.

Vertebrae.— As Anthracodromeus has been said to differ from Ceph-
alerpeton in having hatchet-shaped neural spines and elongate limbs,
these postcranial characters require reexamination. In the type speci-
men of C. ventriarmatum the neural spines are preserved only on the
axis and the seven succeeding vertebrae, so our comparisons must be
restricted to this region. Only in vertebrae 7 through 9 are the profiles
of the neural spines clearly recorded: here the summits of the spines
are not expanded anteroposteriorly. In the more anterior cervicals the
situation is less clear. In the corresponding region of the type specimen

406

Annals of Carnegie Museum

VOL. 52

of A. longipes the neural spines are irregular and their summits ex-
panded, but to a lesser extent than those of the more posterior vertebrae
(compare Figs. 1 and 4 A in Carroll and Baird 1972). Thus the difference,
though real, is not great, and it may be accentuated by the fact that the
first specimen is three-dimensionally preserved while the second is
crushed paper thin.

Forelimb. — In the type of A. longipes the humerus is 8 trunk-centra
long; it articulates closely with a radius and ulna that are at least 5
trunk-centra long. The ends of all these elements are convexly rounded.
In C ventriarmatum an impression of the fleshy forelimb surrounds
the bones, so their relative positions are probably close to natural. The
humerus is 6.7 trunk-centra long; it is separated by a gap from a radius
and ulna that are 4.7 trunk-centra long, and these in turn are widely
separated from the metacarpals. The gaps between the bones are plain
evidence of incomplete ossification. The ends of the limb bones are
concave, indicating that the type of C. ventriarmatum may have been
even more immature than the type individual of A. longipes. This
difference in levels of ossification may account for the differences in
limb proportions. On the other hand, the vertebrae of A. longipes
appear to be more severely compressed than those of C. ventriarmatum.
This difference in crushing would tend to impart greater apparent length
to the centra of A. longipes, giving a falsely low limb to vertebral length
ratio of this long limbed form. These probable differences in levels of
ossification, preservation and compression, make comparisons be-
tween limb to central length ratios difficult.

The foregoing comparisons lead us to conclude that, although there
are similarities between the type species of the nominal genera Ceph-
alerpeton and Anthracodromeus, the evidence to justify a positive state-
ment of synonymy or non-synonymy is either lacking or ambiguous.
This is partly because the Protorothyrididae are inadequately repre-
sented in the fossil record, the morphological variation of particular
taxa is unknown, and the known specimens are difficult to compare as
a result of their fragmentary nature. It is not possible, therefore, to
differentiate the morphological similarities that are indicative of re-
lationships at the familial level from those that are indicative of re-
lationships at generic or specific levels. In such a doubtful case the
taxonomically parsimonious course would be to opt for a tentative
synonymy, while the conservative comsQ would be to retain both generic
names until further evidence is forthcoming. Lacking strong convic-
tions either way, we choose the second course. Rare though these rep-
tiles are, their source localities are still yielding specimens to diligent
collectors; so we may reasonably hope that future finds will clarify the
issue.

1983

Reisz and Baird— Captorhinomorph Reptile Remains

407

In the preceding pages we have demonstrated the presence at Linton
of a reptile that is unquestionably assignable to Cephalerpeton although
it differs from C ventriarmatum in species-level characters. In view of
the possibility that this species may prove to be Cope’s ''Sauropleura''
longipes we refrain from proposing a new specific name, but merely
designate it as Cephalerpeton aff. C ventriarmatum to indicate its
distinctness from the type species.

Revised Diagnoses

Cephalerpeton Moodie, 1912

Type species. — Cephalerpeton ventriarmatum Moodie, 1912

Diagnosis. — Protorothyridid captorhinomorph reptile characterized
by short cheek, dorsally enlarged maxilla, very large marginal teeth
with plicate enamel, reduced dental formula of three premaxillary and
1 6 maxillary teeth, heavily denticulated vomer, and long limbs.

Cephalerpeton ventriarmatum Moodie, 1912

Diagnosis. --M2ind\h\x\2ir dentition relatively unspecialized; mandi-
ble shallower, with 21 or 22 teeth.

//6>nzo«. — Ironstone nodules of Francis Creek Shale overlying Num-
ber 2 (Wilmington or Colchester) Coal, Carbondale Formation, Alle-
gheny Group, Middle Pennsylvanian (early Westphalian D).

Locality. — B2iYiks of Mazon Creek, Grundy County, Illinois.

Cephalerpeton aff. C. ventriarmatum

Diagnosis. — ShovlQV maxillary enclosure of lacrimal duct than in C
ventriarmatum’, dentition more specialized with relatively larger teeth;
dentary deeper, with place for 1 7 teeth.

//6>nz6>A7. — Carbonaceous shale underlying Upper Freeport Coal,
Freeport Formation, Allegheny Group, Middle Pennsylvanian (late
Westphalian D).

Locality. — Didimond Mine (Linton) near Wellsville, west bank near
mouth of Yellow Creek, Saline Township, Jefferson County, Ohio: NE
corner Sect. 13, T9N, R2W.

Anthracodromeus Carroll and Baird, 1972

Type species {monoXypic).—Sauropleura longipes Cope, 1874.

— Protorothyridid captorhinomorph reptile characterized
by short cheek, hatchet-shaped dorsal neural spines, and very long
limbs and feet.

Horizon and locality.— As for Cephalerpeton aff C. ventriarmatum.

408

Annals of Carnegie Museum

VOL. 52

Life Habits

Protorothyridid captorhinomorphs are considered to have been small,
agile, terrestrial animals that fed largely on arthropods (Carroll, 1969;
Carroll and Baird, 1972). Their gracile bodies, long limbs and high
degree of ossification of the skeleton in the better known genera imply
that protorothyridids were agile runners and tree climbers like the
extant small, terrestrial and arboreal lizards. A terrestrial habitat is
also suggested by the occurrence of the Pennsylvanian protorothyridids
Hylonomus and Paleothyris in the hollows of upright Sigillaria stumps
at Joggins and Florence, Nova Scotia. Too much should not be made
of this point, however, for the same trees contain skeletons of evidently
amphibious or aquatic labyrinthodonts— the edopoid Dendrerpeton and
the embolomere Calligenethlon at Joggins, the edopoid Cochleosaurus
and an unnamed embolomere at Florence (Museum of Comparative
Zoology material).

Dietary habits of Paleozoic reptiles are difficult to establish in the
absence of direct evidence, such as stomach contents. Speculations on
preferred diets are generally based upon the known contemporaneous
faunas as well as the skeletal morphology and dental patterns of the
forms in question. It is generally assumed that the protorothyridids
were insectivorous, although comparisons with modem lizards indicate
that they were probably capable of preying on a variety of small ter-
restrial animals, including tetrapods. The marginal dentition seen in
Cephalerpeton is considerably larger and more robust than those of
other protorothyridids or of modem insectivorous lizards. The size and
morphology of its skull and dentition indicate that Cephalerpeton was
able to feed not only on arthropods, but also upon smaller terrestrial
vertebrates, such as some of the lepospondyl amphibians, or young
individuals of protorothyridids. An appropriate menu of insects occurs
in association with the reptiles in the Mazon Creek and Nyfany de-
posits. At Linton, however, insect remains are absent, although the
fauna includes infrequent specimens of more heavily sclerotized ar-
thropods, such as diplopod myriapods (Baird, 1958<2; Hoffman, 1963),
and a pygocephalomorph crustacean (Brooks, 1 962: 1 99). It seems likely
that the robust dentition of Cephalerpeton was able to deal with rela-
tively hard-shelled arthropods of this sort. Myriapods are among the
more common terrestrial arthropods in the Braidwood fauna of Mazon
Creek, and many of them, such as Euphoberia and Acantherpestes,
have spinescent exoskeletons that presumably developed as a defense
against predation (Rolfe, 1980; Hannibal and Feldmann, 1981). The
number of terrestrial predators against which such defenses might have
evolved is limited, but it includes the dissorophid labyrinthodonts
(Carroll, 1964), terrestrial microsaurs such as Tuditanus (Carroll and
Baird, 1968), the smaller pelycosaurs (Reisz, 1972, 1975), and the
protorothyridids.

1983

Reisz and Baird— Captorhinomorph Reptile Remains

409

On the other hand, millipedes are notoriously deliberate pedestrians,
probably requiring little speed or agility to catch. As Anthracodromeus
obviously had both a slender, lightly built body and long limbs it was
able to capture agile invertebrates such as the fleet and nimble cock-
roaches (Blattaria) that abounded in fossil localities of Westphalian
age. Although not preserved at Linton, cockroaches are plentiful in the
Mazon Creek sediments where they constitute 2 1 % of the insect fauna
(Richardson, 1956). It seems most likely that these specialized pro-
torothyridids had evolved their long limbs more as adaptations for
predation than as a means to escape from predation. Their potential
predators, primarily the haptodontine and ophiacodont pelycosaurs,
were evidently much less agile than Anthracodromeus.

In addition to long legs, Anthracodromeus had feet that are remark-
able for their length and areal extent when compared to its slight body
weight. Both manus and pes are almost completely represented in the
type specimen and have been reconstructed by Carroll and Baird (1 972:
Fig. 5). Although in life the feet would have been more compact laterally
and the metapodials less splayed-out than they appear in that recon-
struction, the reptile was doubtless able to scamper over boggy or
muddy surfaces where other tetrapods would have become mired.

Acknowledgments

We are indebted to our colleague Richard Lund, Research Associate of the Carnegie
Museum of Natural History, Pittsburgh, who discovered the crucial specimens. We wish
to thank Kathleen V. Bossy for the latex mold of the British Museum specimen, and the
curators of the institutions cited for providing access to their collections. Suggestions
and criticisms of Malcolm J. Heaton and Robert W. Hook are gratefully acknowledged.
Research was supported by the Natural Science and Engineering Research Council of
Canada, the Jeffries Wyman Fund at Harvard University, and the William Berryman
Scott Fund of Princeton University.

Literature Cited

Baird, D. 1955. Latex micro-molding and latex-plaster molding mixture. Science, 122:
202.

. 1958. The oldest known reptile fauna. Anat. Record, 132:407-408.

. 1958<3. New records of Paleozoic diplopod Myriapoda. J. Paleont., 32:239-

241.

. 1965. Paleozoic lepospondyl amphibians. Amer. Zool., 5:287-294.

Brooks, H. K. 1962. The Paleozoic Eumalacostraca of North America. Bull. Amer.
Paleont., 44:159-338.

Carroll, R. L. 1 964. Early evolution of the dissorophid amphibians. Bull. Mus. Comp.
Zool., Harvard Univ., 131:161-250.

— — . 1969. A Middle Pennsylvanian captorhinomorph, and the interrelationships
of primitive reptiles. J. Paleont., 43:151-170.

Carroll, R. L., and D. Baird. 1968. The Carboniferous amphibian Tuditanus (Eo-
sauravus) and the distinction between microsaurs and reptiles. Amer. Mus. Novit.,
2337:1-50.

. 1 972. Carboniferous stem-reptiles of the family Romeriidae. Bull. Mus. Comp.

Zool., Harvard Univ., 143:353-407.

410

Annals of Carnegie Museum

VOL. 52

Clark, J. B., and R. L. Carroll. 1973. Romeriid reptiles from the Lower Permian.

Bull. Mus. Comp. ZooL, Harvard Univ., 144:353-407.

Cope, E. D. 1874. Supplement to the extinct Batrachia and Reptilia of North America.
I. Catalogue of the air breathing Vertebrata from the Coal Measures of Linton, Ohio.
Trans. Amer. Phil. Soc., (N.S.), 15:261-278.

Currie, P. J. 1979. The osteology of haptodontine sphenacodonts (Reptilia: Pelyco-
sauria). Palaeontographica, (A) 164:130-168.

Gaffney, E. S., and M. C. McKenna. 1979. A Late Permian captorhinid from Rho-
desia. Amer. Mus. Novit., 2688:1-15.

Gregory, J. T. 1948. The structure of Cephalerpeton and the affinities of the Micro-
sauria. Amer. J. Sci., 246:550-568.

. 1950. Tetrapods of the Pennsylvanian nodules from Mazon Creek, Illinois.

Amer. J. Sci., 248:833-873.

Hannibal, J. T., and R. M. Feldmann. 1981. Systematics and functional morphology
of oniscomorph millipedes (Arthropoda: Diplopoda) from the Carboniferous of
North America. J. Paleont., 55:730-746.

Heaton, M. J. 1979. Cranial anatomy of primitive captorhinid reptiles from the Late
Pennsylvanian and Early Permian, Oklahoma and Texas. Oklahoma Geol. Surv.
Bull., 127:i-iv+ 1-84.

Hook, R. 1981. The Pennsylvanian tetrapod fauna of Linton, Ohio. Ohio J. Sci., 81:
41.

Hoffman, R. L. 1963. New genera and species of Upper Paleozoic Diplopoda. J.
Paleont., 37:167-174.

Konzhukova, E. D. 1956. (The Intyn Lower Permian fauna of the northern Ural
region). Trav. Inst. Paleont., Acad. Sci. URSS, 62:5-50.

KuTTY, T. S. 1972. Permian reptilian fauna from India. Nature, 237:462-463.
Milner, A. R. 1980. The tetrapod assemblage from Nyfany, Czechoslovakia. Pp. 439-
496, in The terrestrial environment and the origin of land vertebrates (A. L. Panchen,
ed.), Systematics Assoc. Special, Academic Press, 15:1-633.

Moodie, R. L. 1912. The Pennsylvanic Amphibia of the Mazon Creek, Illinois, shales.
Kansas Univ. Sci. Bull., 16:323-359.

Olson, E. C. 1952. The evolution of a Permian vertebrate chronofauna. Evolution,
6:181-196.

. 1954. Fauna of the Vale and Choza: 9, Captorhinomorpha. Fieldiana: Geol.,

10:211-218.

. 1962. Late Permian terrestrial vertebrates, U.S.A. and U.S.S.R. Trans. Amer.

Phil. Soc., 52:1-224.

Olson, E. C., and J. R. Beerbower. 1953. The San Angelo Formation, Permian of
Texas, and its vertebrates. J. Geol., 61:389-423.

Price, L. I. 1937. Two new cotylosaurs from the Permian of Texas. Proc. New England
Zool. Club, 16:97-102.

Rayner, D. H. 1971. Data on the environment and preservation of Late Paleozoic
tetrapods. Proc. Yorkshire Geol. Soc., 38:437-495.

Reisz, R. R. 1972. Pelycosaurian reptiles from the Middle Pennsylvanian of North
America. Bull. Mus. Comp. Zool., Harvard Univ., 144:27-62.

. 1975. Pennsylvanian pelycosaurs from Linton, Ohio, and Nyfany, Czechoslo-
vakia. J. Paleont., 49:522-527.

— . 1980. A protorothyridid captorhinomorph reptile from the Lower Permian of

Oklahoma. Life Sci. Contrib., Royal Ontario Mus., 121:1-17.

Richardson, E. S., Jr. 1956. Pennsylvanian invertebrates of the Mazon Creek area,
Illinois. Fieldiana: Geol., 12:13-56.

Rolfe, W. D. I. 1980. Early invertebrate terrestrial faunas. Pp. 1 17-157 in The ter-

1983

Reisz and Baird— Captorhinomorph Reptile Remains

411

restrial environment and the origin of land vertebrates (A. L. Panchen, ed.), Sys-
tematics Assoc. Special, Academic Press, 15:1 ”6 33.

Steen, M. C. 1931. The British Museum collection of Amphibia from the Middle Coal
Measures of Linton, Ohio. Proc. Zool. Soc. London, 1930:849-891.

Taquet, P. 1969. Premiere decouverte en Afrique d’un reptile captorhinomorphe
(Cotylosaurien). C. R. Acad. Sci., Paris, 268:779-781.

Westoll, T. S. 1944. The Haplolepidae, a new family of Late Carboniferous bony
fishes. Bull. Amer. Mus. Nat. Hist., 83:1-121.

Back issues of many Annals of Carnegie Museum articles are
available, and a few early complete volumes and parts are listed
at half price. Orders and inquiries should be addressed to:
Publications Secretary, Carnegie Museum, 4400 Forbes Avenue,
Pittsburgh, Pa. 15213.

r

--S'. „ "S

i«, v^'fis

::'»'ri„'

:' WM--

■ , ►,,. /■ .Vr ■ '.,

yJ

,. „ I 'f'.i .1 . !• N ,. . ■■■,

^ - ^f ,, ?;; ■:/:

:>

.-V . . ,. ■ ■ ’'A ' /

v wti M A
■'''• A '

‘ ;. 'I •:

‘'i-^ ■" wi

aV'l- ■' ' ' ' ': ':

»*;'t'V"',';AA ‘^' T!'"'; '/■ :v^;:

,^‘. . „, j>)r'i'.'‘''r,iV, ■ ' :,K.i.' ' V 'I'

-. , "vj'..V,«, ’r ^'■, . >1) ■ - 'v'.C

, .,. V.'.

■ - 5'

RARIES SMITHSONIAN INSTITUTION NOliniliSNl NVINOSHilWS SBIdVHS

NVmOSHilWS S3IHVH8n LIBRARIES SMITHSONIAN INSTITUTIC

RARIES SMITHSONIAN INSTITUTION NOliniliSNl NVINOSHifINS SBIBVHB

RAR I ES^^SMiTHSONIAN^INSTITUTlON^^NOliniliSNl^NVINOSHillNS^Sa 1 BVBS

iLi

IT

<

cc

m

CO

70

>

TO

m

If)

RARIES SMITHSONIAN INSTITUTION NOliHiliSNI

NVINOSHimS S3IBVBa

CO

iniliSNi

X
U)

o

X

s >

NVINOSHiliAls‘^S3 I UVU 3 Ilf

LIBRARIES SMITHSONIAN

U)
yj

INSTITUTII

liSNI^NVINOSHilWS^SH lavyan^LIBRARt ES^SMITHSONIAN^INSTITUTION^

^ ^ m ^

R I ES SMITHS0NIAN~lNSTITUTI0N‘”N0liniliSNl“NVIN0SHilWs'”s3 I aVM 8 IT

^ ^ Z \ cn z cn

Q ^ s s (li. ill s

> 2 3, ■ 2 >

[iSNI_NVIN0SHill/\IS^S3 ! ^ VH 8 ll^L! B RAR I ES^SMITHSONIAN^INSTITUTION ^
Z \ ^ ^ ^ in zz (n

in ..^ ^ /A 4jJ ^ >^'a iOv UJ

cc li,

< loi

a:

2 _l z

R I ES SMITHSONIAN INSTITUTION NOIlfliliSNI NVINOSHillNS SBlUVHail

— in ^ m \ z ■ —

liSNlMVINOSHlIlAISSaiaVaan LIBRARIES SMITHSONIAN INSTITUTION

^ ^ S5 z CO ^

^ 8 (m^ 5 s i /w^ o

^^7 O

W E

^ > _

RIES SMITHSONIAN_INSTITUTION NOlinillSNI NVIN0SHlll^S^S3 I 8 VU 8 11

“ .-r— ^ ^ ... </>

o __ x^iiis^ O

ISNl^NVINOSHillNS S3 I UVa 8 li “"lI B RAR I ES^SM1THS0NIAN“^INSTITUTI0N

' V Z r- 2 f

W
33
•i>

PI I w* '• •• -

n ES^SMITHSONIAN INSTITUTI0N‘”N0liniaSNl“NVIN0SHilWS*”s3 1 aVB 8 IT

S -C^S^Sx < '\x s i

X1C^S^» o X teJE ^ O

^ I*' i >•

m^S3 1 a va a n^Li B rar i es^smithsonian ^institution